CHAPTER 1 Fundamentals of testing
n this chapter, we will introduce you to the fundamentals of testing: why testing is needed; its limitations, objectives and purpose; the principles behind testing; the process that testers follow; and some of the psychological factors that testers must consider in their work. By reading this chapter you'll gain an understanding of the fundamentals of testing and be able to describe those fundamentals.
1.1 WHY IS TESTING NECESSARY?
1 Describe, with examples, the way in which a defect in software can cause harm to a person, to the environment or to a company. (K2)
2 Distinguish between the root cause of a defect and its effects. (K2)
3 Give reasons why testing is necessary by giving examples. (K2)
4 Describe why testing is part of quality assurance and give examples of how testing contributes to higher quality. (K2)
5 Recall the terms 'mistake', 'defect', 'fault', 'failure' and the correspon ding terms 'error' and 'bug'. (Kl)
6 Explain the fundamental principles in testing. (K2)
1.1.1 Introduction
In this section, we're going to kick off the book with a discussion on why testing matters. We'll describe and illustrate how software defects or bugs can cause problems for people, the environment or a company. We'll draw important dis-tinctions between defects, their root causes and their effects. We'll explain why testing is necessary to find these defects, how testing promotes quality, and how testing fits into quality assurance. In this section, we will also introduce some fundamental principles of testing.
As we go through this section, watch for the Syllabus terms bug, defect, error, failure, fault, mistake, quality, risk, software, testing and exhaustive testing. You'll find these terms defined in the glossary.
You may be asking 'what is testing?' and we'll look more closely at the defi-nition of testing in Section 1.2. For the moment, let's adopt a simple everyday-life usage: 'when we are testing something we are checking whether it is OK'. We'll need to refine that when we define software testing later on. Let's start by considering why testing is needed. Testing is necessary because we all make mis-takes. Some of those mistakes are unimportant, but some of them are expensive or dangerous. We need to check everything and anything we produce because things can always go wrong - humans make mistakes all the time - it is what we do best!
Because we should assume our work contains mistakes, we all need to check our own work. However, some mistakes come from bad assumptions and blind spots, so we might make the same mistakes when we check our own work as we made when we did it. So we may not notice the flaws in what we have done. Ideally, we should get someone else to check our work - another person is more likely to spot the flaws.
In this book, we'll explore the implications of these two simple paragraphs again and again. Does it matter if there are mistakes in what we do? Does it matter if we don't find some of those flaws? We know that in ordinary life, some of our mistakes do not matter, and some are very important. It is the same with software systems. We need to know whether a particular error is likely to cause problems. To help us think about this, we need to consider the context within which we use different software systems.
1.1.2 Software systems context
Testing Principle - Testing is context dependent
Testing is done differently in different contexts. For example, safety-critical software is tested differently from an e-commerce site.
These days, almost everyone is aware of software systems. We encounter them in our homes, at work, while shopping, and because of mass-communication systems. More and more, they are part of our lives. We use software in day-to-day business applications such as banking and in consumer products such as cars and washing machines. However, most people have had an experience with software that did not work as expected: an error on a bill, a delay when waiting for a credit card to process and a website that did not load correctly are common examples of problems that may happen because of software problems. Not all software systems carry the same level of risk and not all problems have the same impact when they occur. A risk is something that has not hap-pened yet and it may never happen; it is a potential problem. We are concerned about these potential problems because, if one of them did happen, we'd feel a negative impact. When we discuss risks, we need to consider how likely it is that the problem would occur and the impact if it happens. For example, whenever we cross the road, there is some risk that we'll be injured by a car. The likeli-hood depends on factors such as how much traffic is on the road, whether there is a safe crossing place, how well we can see, and how fast we can cross. The impact depends on how fast the car is going, whether we are wearing protective gear, our age and our health. The risk for a particular person can be worked out and therefore the best road-crossing strategy.
Some of the problems we encounter when using software are quite trivial, but others can be costly and damaging - with loss of money, time or business reputation - and even may result in injury or death. For example, suppose a user interface has typographical defects. Does this matter? It may be trivial, but it could have a significant effect, depending on the website and the defect:
• If my personal family-tree website has my maternal grandmother's maiden name spelt wrong, my mother gets annoyed and I have to put up with some family teasing, but I can fix it easily and only the family see it (probably).
• If the company website has some spelling mistakes in the text, potential cus tomers may be put off the company as it looks unprofessional.
• If a software program miscalculates pesticide application quantities, the effect could be very significant: suppose a decimal point is wrongly placed so that the application rate is 10 times too large. The farmer or gardener uses more pesticide than needed, which raises his costs, has environmental impacts on wildlife and water supplies and has health and safety impact for the farmer, gardener, family and workforce, livestock and pets. There may also be consequent loss of trust in and business for the company and possi ble legal costs and fines for causing the environmental and health problems.
1.1.3 Causes of software defects
Why is it that software systems sometimes don't work correctly? We know that people make mistakes - we are fallible.
If someone makes an error or mistake in using the software, this may lead directly to a problem - the software is used incorrectly and so does not behave as we expected. However, people also design and build the software and they can make mistakes during the design and build. These mistakes mean that there are flaws in the software itself. These are called defects or sometimes bugs or faults. Remember, the software is not just the code; check the definition of soft-ware again to remind yourself.
When the software code has been built, it is executed and then any defects may cause the system to fail to do what it should do (or do something it shouldn't), causing a failure. Not all defects result in failures; some stay dormant in the code and we may never notice them.
Do our mistakes matter?
Let's think about the consequences of mistakes. We agree that any human
being, programmers and testers included, can make an error. These errors may
produce defects in the software code or system, or in a document. If a defect in
code is executed, the system may experience a failure. So the mistakes we make
matter partly because they have consequences for the products for which we are
responsible.
Our mistakes are also important because software systems and projects are complicated. Many interim and final products are built during a project, and people will almost certainly make mistakes and errors in all the activities of the build. Some of these are found and removed by the authors of the work, but it is difficult for people to find their own mistakes while building a product. Defects in software, systems or documents may result in failures, but not all
defects do cause failures. We could argue that if a mistake does not lead to a defect or a defect does not lead to a failure, then it is not of any importance -we may not even know we've made an error.
Our fallibility is compounded when we lack experience, don't have the right information, misunderstand, or if we are careless, tired or under time pressure. All these factors affect our ability to make sensible decisions - our brains either don't have the information or cannot process it quickly enough.
Additionally, we are more likely to make errors when dealing with perplex-ing technical or business problems, complex business processes, code or infra-structure, changing technologies, or many system interactions. This is because our brains can only deal with a reasonable amount of complexity or change -when asked to deal with more our brains may not process the information we have correctly.
It is not just defects that give rise to failure. Failures can also be caused by environmental conditions as well: for example, a radiation burst, a strong mag-netic field, electronic fields, or pollution could cause faults in hardware or firmware. Those faults might prevent or change the execution of software. Failures may also arise because of human error in interacting with the software, perhaps a wrong input value being entered or an output being misinterpreted. Finally, failures may also be caused by someone deliberately trying to cause a failure in a system - malicious damage.
When we think about what might go wrong we have to consider defects and failures arising from:
• errors in the specification, design and implementation of the software and system;
• errors in use of the system;
• environmental conditions;
• intentional damage;
• potential consequences of earlier errors, intentional damage, defects and failures.
When do defects arise?
In Figure 1.1 we can see how defects may arise in four requirements for a
product.
We can see that requirement 1 is implemented correctly - we understood the customer's requirement, designed correctly to meet that requirement, built cor-rectly to meet the design, and so deliver that requirement with the right attrib-utes: functionally, it does what it is supposed to do and it also has the right non-functional attributes, so it is fast enough, easy to understand and so on.
With the other requirements, errors have been made at different stages. Requirement 2 is fine until the software is coded, when we make some mistakes and introduce defects. Probably, these are easily spotted and corrected during testing, because we can see the product does not meet its design specification.
The defects introduced in requirement 3 are harder to deal with; we built exactly what we were told to but unfortunately the designer made some mis-takes so there are defects in the design. Unless we check against the require-ments definition, we will not spot those defects during testing. When we do notice them they will be hard to fix because design changes will be required.
The defects in requirement 4 were introduced during the definition of the requirements; the product has been designed and built to meet that flawed requirements definition. If we test the product meets its requirements and design, it will pass its tests but may be rejected by the user or customer. Defects reported by the customer in acceptance test or live use can be very costly. Unfortunately, requirements and design defects are not rare; assessments of thousands of projects have shown that defects introduced during requirements and design make up close to half of the total number of defects [Jones].
What is the cost of defects?
As well as considering the impact of failures arising from defects we have not found, we need to consider the impact of when we find those defects. The cost of finding and fixing defects rises considerably across the life cycle; think of the old English proverb 'a stitch in time saves nine'. This means that if you mend a tear in your sleeve now while it is small, it's easy to mend, but if you leave it, it will get worse and need more stitches to mend it.
If we relate the scenarios mentioned previously to Figure 1.2, we see that, if an error is made and the consequent defect is detected in the requirements at the specification stage, then it is relatively cheap to find and fix. The observa-tion of increasing defect-removal costs in software traces back to [Boehm]. You'll find evidence for the economics of testing and other quality assurance activities in [Gilb], [Black 2001] or [Black 2004]. The specification can be cor-rected and re-issued. Similarly if an error is made and the consequent defect detected in the design at the design stage then the design can be corrected and re-issued with relatively little expense. The same applies for construction. If
however a defect is introduced in the requirement specification and it is not detected until acceptance testing or even once the system has been imple-mented then it will be much more expensive to fix. This is because rework will be needed in the specification and design before changes can be made in con-struction; because one defect in the requirements may well propagate into several places in the design and code; and because all the testing work done-to that point will need to be repeated in order to reach the confidence level in the software that we require.
It is quite often the case that defects detected at a very late stage, depending on how serious they are, are not corrected because the cost of doing so is too expensive. Also, if the software is delivered and meets an agreed specification, it sometimes still won't be accepted if the specification was wrong. The project team may have delivered exactly what they were asked to deliver, but it is not what the users wanted. This can lead to users being unhappy with the system that is finally delivered. In some cases, where the defect is too serious, the system may have to be de-installed completely.
1.1.4 Role of testing in software development, maintenance and operations
We have seen that human errors can cause a defect or fault to be introduced at any stage within the software development life cycle and, depending upon the consequences of the mistake, the results can be trivial or catastrophic. Rigorous testing is necessary during development and maintenance to identify defects, in order to reduce failures in the operational environment and increase the quality of the operational system. This includes looking for places in the user interface where a user might make a mistake in input of data or in the interpretation of the output, and looking for potential weak points for intentional and malicious attack. Executing tests helps us move towards improved quality of product and service, but that is just one of the verification and validation methods applied to products. Processes are also checked, for example by audit. A variety of methods may be used to check work, some of which are done by the author of the work and some by others to get an independent view.
We may also be required to carry out software testing to meet contractual or legal requirements, or industry-specific standards. These standards may specify what type of techniques we must use, or the percentage of the software code that must be exercised. It may be a surprise to learn that we don't always test all the code; that would be too expensive compared with the risk we are trying deal with. However - as we'd expect - the higher the risk associated with the indus-try using the software, the more likely it is that a standard for testing will exist. The avionics, motor, medical and pharmaceutical industries all have standards covering the testing of software. For example, the US Federal Aviation Administration's DO-178B standard [RTCA/DO-178B] has requirements for test coverage.
1.1.5 Testing and quality
Testing helps us to measure the quality of software in terms of the number of defects found, the tests run, and the system covered by the tests. We can do this for both the functional attributes of the software (for example, printing a report correctly) and for the non-functional software requirements and characteristics (for example, printing a report quickly enough). Non-functional characteristics are covered in Chapter 2. Testing can give confidence in the quality of the soft-ware if it finds few or no defects, provided we are happy that the testing is suf-ficiently rigorous. Of course, a poor test may uncover few defects and leave us with a false sense of security. A well-designed test will uncover defects if they are present and so, if such a test passes, we will rightly be more confident in the software and be able to assert that the overall level of risk of using the system has been reduced. When testing does find defects, the quality of the software system increases when those defects are fixed, provided the fixes are carried out properly.
What is quality?
Projects aim to deliver software to specification. For the project to deliver what the customer needs requires a correct specification. Additionally, the delivered system must meet the specification. This is known as validation ('is this the right specification?') and verification ('is the system correct to spec-ification?'). Of course, as well as wanting the right software system built cor-rectly, the customer wants the project to be within budget and timescale - it should arrive when they need it and not cost too much.
The ISTQB glossary definition covers not just the specified requirements but also user and customer needs and expectations. It is important that the project team, the customers and any other project stakeholders set and agree expecta-tions. We need to understand what the customers understand by quality and what their expectations are. What we as software developers and testers may see as quality - that the software meets its defined specification, is technically excellent and has few bugs in it - may not provide a quality solution for our cus-tomers. Furthermore, if our customers find they have spent more money than they wanted or that the software doesn't help them carry out their tasks, they won't be impressed by the technical excellence of the solution. If the customer wants a cheap car for a 'run-about' and has a small budget then an expensive
sports car or a military tank are not quality solutions, however well built they are.
To help you compare different people's expectations, Table 1.1 summarizes and explains quality viewpoints and expectations using 'producing and buying tomatoes' as an analogy for 'producing and buying software'. You'll see as you look through the table that the approach to testing would be quite different depending on which viewpoint we favor [Trienekens], [Evans].
In addition to understanding what quality feels and looks like to customers, users, and other stakeholders, it helps to have some quality attributes to measure quality against, particularly to aid the first, product-based, viewpoint in the table. These attributes or characteristics can serve as a framework or checklists for areas to consider coverage. One such set of quality attributes can
TABLE 1.1 Viewpoints of expectations and quality
Viewpoint Software Tomatoes
Quality is measured by looking at the We will measure the attributes of the The tomatoes are the right size
attributes of the product. software, e.g. its reliability in terms of and shape for packing for the
mean time between failures (MBTF), supermarket. The tomatoes
and release when they reach a have a good taste and color, specified level e.g. MTBF of 12 hours.
Quality is fitness for use. Quality can We will ask the users whether they The tomatoes are right for our have subjective aspects and not just can carry out their tasks; if they are recipe,
quantitative aspects. satisfied that they can we will release
the software.
Quality is based on good manufacturing We will use a recognized software The tomatoes are organically
processes, and meeting defined development process. We will only farmed. The tomatoes have no
requirements. It is measured by testing, release the software if there are fewer blemishes and no pest
inspection, and analysis of faults and than five outstanding high-priority damage, failures. defects once the planned tests are
complete.
Expectation of value for money. We have time-boxed the testing to The tomatoes have a good
affordability, and a value-based trade-off two weeks to stay in the project shelf life. The tomatoes are
between time, effort and cost aspects. budget. cheap or good value for
We can afford to buy this software and money,
we expect a return on investment.
Transcendent feelings - this is about the We like this software! It is fun and it's We get our tomatoes from a feelings of an individual or group of the latest thing! So what if it has a small local farm and we get on
individuals towards a product or a few problems? We want to use it so well with the growers,
supplier. anyway...
We really enjoy working with this
software team. So, there were a few
problems - they sorted them out
really quickly - we trust them.
be found in the ISO 9126 standard. This hierarchy of characteristics and sub-characteristics of quality is discussed in Chapter 2.
What is root cause analysis?
When we detect failures, we might try to track them back to their root cause, the real reason that they happened. There are several ways of carrying out root cause analysis, often involving a group brainstorming ideas and discussing them, so you may see different techniques in different organizations. If you are inter-ested in using root cause analysis in your work, you'll find simple techniques described in [Evans], [TQMI] and [Robson]. For example, suppose an organi-zation has a problem with printing repeatedly failing. Some IT maintenance folk get together to examine the problem and they start by brainstorming all the possible causes of the failures. Then they group them into categories they have chosen, and see if there are common underlying or root causes. Some of the obvious causes they discover might be:
• Printer runs out of supplies (ink or paper).
• Printer driver software fails.
• Printer room is too hot for the printer and it seizes up.
These are the immediate causes. If we look at one of them - 'Printer runs out of supplies (ink or paper)' - it may happen because:
• No-one is responsible for checking the paper and ink levels in the printer; possible root cause: no process for checking printer ink/paper levels before use.
• Some staff don't know how to change the ink cartridges; possible root cause: staff not trained or given instructions in looking after the printers.
• There is no supply of replacement cartridges or paper; possible root cause: no process for stock control and ordering.
If your testing is confined to software, you might look at these and say, 'These are not software problems, so they don't concern us!' So, as software testers we might confine ourselves to reporting the printer driver failure. However, our remit as testers may be beyond the software; we might have a remit to look at a whole system including hardware and firmware. Additionally, even if our remit is software, we might want to consider how software might help people prevent or resolve problems; we may look beyond this view. The software could provide a user interface which helps the user anticipate when paper or ink is getting low. It could provide simple step-by-step instructions to help the users change the cartridges or replenish paper. It could provide a high temperature warning so that the environment can be managed. As testers, we want not just to think and report on defects but, with the rest of the project team, think about any potential causes of failures.
We use testing to help us find faults and (potential) failures during software development, maintenance and operations. We do this to help reduce the risk of failures occurring in an operational environment - in other words once the system is being used - and to contribute to the quality of the software system. However, whilst we need to think about and report on a wide variety of defects and failures, not all get fixed. Programmers and others may correct defects
before we release the system for operational use, but it may be more sensible to work around the failure. Fixing a defect has some chance of introducing another defect or of being done incorrectly or incompletely. This is especially true if we are fixing a defect under pressure. For this reason, projects will take a view sometimes that they will defer fixing a fault. This does not mean that the tester who has found the problems has wasted time. It is useful to know that there is a problem and we can help the system users work around and avoid it.
The more rigorous our testing, the more defects we'll find. But you'll see in Chapters 3 and 4, when we look at techniques for testing, that rigorous testing does not necessarily mean more testing; what we want to do is testing that finds defects - a small number of well-placed, targeted tests may be more rigorous than a large number of badly focused tests.
We saw earlier that one strategy for dealing with errors, faults and failures is to try to prevent them, and we looked at identifying the causes of defects and failures. When we start a new project, it is worth learning from the problems encountered in previous projects or in the production software. Understanding the root causes of defects is an important aspect of quality assurance activities, and testing contributes by helping us to identify defects as early as possible before the software is in use. As testers, we are also interested in looking at defects found in other projects, so that we can improve our processes. Process improvements should prevent those defects recurring and, as a consequence, improve the quality of future systems. Organizations should consider testing as part of a larger quality assurance strategy, which includes other activities (e.g., development standards, training and root cause analysis).
1.1.6 How much testing is enough?
Testing Principle - Exhaustive testing is impossible
Testing everything (all combinations of inputs and preconditions) is not feasible except for trivial cases. Instead of exhaustive testing, we use risks and priorities to focus testing efforts.
We've seen that testing helps us find defects and improve software quality. How much testing should we do? We have a choice: test everything, test nothing or test some of the software. Now, your immediate response to that may well be to say, 'Everything must be tested'. We don't want to use software that has not been completely tested, do we? This implies that we must exercise every aspect of a software system during testing. What we need to consider is whether we must, or even can, test completely.
Let's look at how much testing we'd need to do to be able to test exhaus-tively. How many tests would you need to do to completely test a one-digit numeric field? The immediate question is, 'What you mean by test completely?' There are 10 possible valid numeric values but as well as the valid values we need to ensure that all the invalid values are rejected. There are 26 uppercase alpha characters, 26 lower case, at least 6 special and punctuation characters as well as a blank value. So there would be at least 68 tests for this example of a one-digit field.
This problem just gets worse as we look at more realistic examples. In prac-tice, systems have more than one input field with the fields being of varying sizes. These tests would be alongside others such as running the tests in differ-
Section 2 What is testing? 11
ent environments. If we take an example where one screen has 15 input fields, each having 5 possible values, then to test all of the valid input value combina-tions you would need 30 517 578 125 (515) tests! It is unlikely that the project timescales would allow for this number of tests.
Testing our one-digit field with values 2, 3 and 4 makes our tests more thor-ough, but it does not give us more information than if we had just tested with the value 3.
Pressures on a project include time and budget as well as pressure to deliver a technical solution that meets the customers' needs. Customers and project managers will want to spend an amount on testing that provides a return on investment for them. This return on investment includes prevent-ing failures after release that are costly. Testing completely - even if that is what customers and project managers ask for - is simply not what they can afford.
Instead we need a test approach which provides the right amount of testing for this project, these customers (and other stakeholders) and this software. We do this by aligning the testing we do with the risks for the customers, the stake-holders, the project and the software. Assessing and managing risk is one of the most important activities in any project, and is a key activity and reason for testing. Deciding how much testing is enough should take account of the level of risk, including technical and business risks related to the product and project constraints such as time and budget.
We carry out a risk assessment to decide how much testing to do. We can then vary the testing effort based on the level of risk in different areas. Additionally, testing should provide sufficient information to stakeholders to make informed decisions about the release of the software or system we're testing, for the next development step or handover to customers. The effort put into the quality assurance and testing activities needs to be tai lored to the risks and costs associated with the project. Because of the limits in the budget, the time, and in testing we need to decide how we will focus our testing, based on the risks. We'll look at risk assessment in Chapter 5.
1.2 WHAT IS TESTING?
Syllabus learning objectives for 1.2 What is testing?
1 Recall the common objectives of testing. (Kl)
2 Describe the purpose of testing in software development, maintenance and operations as a means to find defects, provide confidence and infor mation, and prevent defects. (K2)
In this section, we will review the common objectives of testing. We'll explain how testing helps us to find defects, provide confidence and information, and prevent defects. We will also introduce additional fundamental principles of testing.
As you read this section, you'll encounter the terms code, debugging, development of software, requirement, review, test basis, test case, testing and test objective.
1.2.1 The driving test - an analogy for software testing
We have spent some time describing why we need to test, but we have not dis-cussed what testing is. What do we mean by the word testing? We use the words test and testing in everyday life and earlier we said testing could be described as 'checking the software is OK'. That is not a detailed enough definition to help us understand software testing. Let's use an analogy to help us: driving tests. In a driving test, the examiner critically assesses the candidate's driving, noting every mistake, large or small, made by the driver under test. The examiner takes the driver through a route which tests many possible driving activities, such as road junctions of different types, control and maneuvering of the car, ability to stop safely in an emergency, and awareness of the road, other road users and hazards. Some of the activities must be tested. For example, in the UK, an emergency stop test is always carried out, with the examiner simulating the moment of emergency by hitting the dashboard at which point the driver must stop the car quickly, safely and without loss of control. At the end of the test, the examiner makes a judgment about the driver's performance. Has the driver passed the test or failed? The examiner bases the judgment on the number and severity of the failures identified, and also whether the driver has been able to meet the driving requirements. A single severe fault is enough to fail the whole test, but a small number of minor faults might still mean the test is passed. Many minor faults would reduce the confidence of the examiner in the quality —of the driving to the point where the driver cannot pass. The format of the driving test and the conduct of the examiner are worth considering:
• The test is planned and prepared for. In advance of the test, the examiner has planned a series of routes which cover the key driving activities to allow a thorough assessment of the driver's performance. The drivers under test do not know what route they will be asked to take in advance, although they know the requirements of the test.
• The test has known goals - assessing whether the driver is sufficiently safe to be allowed to drive by themselves without an instructor, without endanger ing themselves or others. There are clear pass/fail criteria, based on the number and severity of faults, but the confidence of the examiner in the driving is also taken into account.
• The test is therefore carried out to show that the driver satisfies the require ments for driving and to demonstrate that they are fit to drive. The examiner looks for faults in the driving. The time for the test is limited, so it is not a complete test of the driver's abilities, but it is representative and allows the examiner to make a risk-based decision about the driver. All the drivers are tested in an equivalent way and the examiner is neutral and objective. The examiner will log factual observations which enable a risk assessment to be made about the driving. Based on this, a driver who passes will be given a form enabling him to apply for a full driving license. A driver who fails will get a report with a list of faults and areas to improve before retaking the test.
• As well as observing the driver actually driving, the examiner will ask questions or the driver will take a written exam to check their under standing of the rules of the road, road signs, and what to do in various traffic situations.
1.2.2 Defining software testing
With that analogy in mind, let's look at the ISTQB definition of software testing.
Let's break the definition down into parts; the definition has some key phrases to remember. The definition starts with a description of testing as a process and then lists some objectives of the test process. First, let's look at testing as a process:
• Process - Testing is a process rather than a single activity - there are a series of activities involved.
• All life cycle activities - Chapter 2 looks at testing as a process that takes place throughout the software development life cycle. We saw earlier that the later in the life cycle we find bugs, the more expensive they are to fix. If we can find and fix requirements defects at the requirements stage, that must make commercial sense. We'll build the right software, correctly and at a lower cost overall. So, the thought process of design ing tests early in the life cycle can help to prevent defects from being introduced into code. We sometimes refer to this as 'verifying the test basis via the test design'. The test basis includes documents such as the requirements and design specifications. You'll see how to do this in Chapter 4.
• Both static and dynamic - We'll see in Chapter 3 that as well as tests where the software code is executed to demonstrate the results of running tests (often called dynamic testing) we can also test and find defects without exe cuting code. This is called static testing. This testing includes reviewing of documents (including source code) and static analysis. This is a useful and cost effective way of testing.
• Planning - Activities take place before and after test execution. We need to manage the testing; for example, we plan what we want to do; we control the test activities; we report on testing progress and the status of the software under test; and we finalize or close testing when a phase completes. Chapter 5 covers these test management activities.
• Preparation - We need to choose what testing we'll do, by selecting test con ditions and designing test cases. Chapter 4 covers the test design activities.
• Evaluation - As well as executing the tests, we must check the results and evaluate the software under test and the completion criteria, which help us decide whether we have finished testing and whether the software product has passed the tests.
• Software products and related work products - We don't just test code. We test the requirements and design specifications, and we test related documents such as operation, user and training material. Static and dynamic testing are both needed to cover the range of products we need to test.
The second part of the definition covers the some of the objectives for testing -the reasons why we do it:
• Determine that (software products) satisfy specified requirements - Some of the testing we do is focused on checking products against the specification for the product; for example we review the design to see if it meets require ments, and then we might execute the code to check that it meets the design. If the product meets its specification, we can provide that information to help stakeholders judge the quality of the product and decide whether it is ready for use.
• Demonstrate that (software products) are fit for purpose - This is slightly different to the point above; after all the specified requirements might be wrong or incomplete. 'Fit for purpose' looks at whether the software does enough to help the users to carry out their tasks; we look at whether the soft ware does what the user might reasonably expect. For example, we might look at who might purchase or use the software, and check that it does do what they expect; this might lead us to add a review of the marketing mate rial to our static tests, to check that expectations of the software are properly set. One way of judging the quality of a product is by how fit it is for its purpose.
• Detect defects - We most often think of software testing as a means of detecting faults or defects that in operational use will cause failures. Finding the defects helps us understand the risks associated with putting the software into operational use, and fixing the defects improves the quality of the prod ucts. However, identifying defects has another benefit. With root cause analysis, they also help us improve the development processes and make fewer mistakes in future work.
This is a suitable definition of testing for any level of testing, from compo-nent testing through to acceptance testing, provided that we remember to take the varying objectives of these different levels of testing into account. (In Chapter 2 we'll cover the different test levels, their objectives, and how they fit into the software development life cycle.)
We can clearly see now why the common perception of testing (that it only consists of running tests, i.e. executing the software) is not complete. This is one of the testing activities, but not all of the testing process.
1.2.3 Software test and driving test compared
We can see that the software test is very like a driving test in many ways, although of course it is not a perfect analogy! The driving examiner becomes the software tester. The driver being examined becomes the system or software under test, and you'll see as we go through this book that the same approach broadly holds.
• Planning and preparation - Both the examiner and the tester need a plan of action and need to prepare for the test, which is not exhaustive, but is repre sentative and allows risk-based decisions about the outcome.
• Static and dynamic - Both dynamic (driving the car or executing the soft ware) and static (questions to the driver or a review of the software) tests are useful.
• Evaluation - The examiner and the tester must make an objective evaluation, log the test outcome and report factual observations about the test.
• Determine that they satisfy specified requirements - The examiner and tester both check against requirements to carry out particular tasks successfully.
• Demonstrate that they are fit for purpose - The examiner and the tester are not evaluating for perfection but for meeting sufficient of the attributes required to pass the test.
• Detect defects - The examiner and tester both look for and log faults.
Let's think a little more about planning. Because time is limited, in order to make a representative route that would provide a sufficiently good test, both software testers and driving examiners decide in advance on the route they will take. It is not possible to carry out the driving test and make decisions about where to ask the driver to go next on the spur of moment. If the examiner did that, they might run out of time and have to return to the test center without having observed all the necessary maneuvers. The driver will still want a pass/fail report. In the same way, if we embark on testing a software system without a plan of action, we are very likely to run out of time before we know whether we have done enough testing. We'll see that good testers always have a plan of action. In some cases, we use a lightweight outline providing the goals and general direction of the test, allowing the testers to vary the test during execution. In other cases, we use detailed scripts showing the steps in the test route and documenting exactly what the tester should expect to happen as each step. Whichever approach the tester takes, there will be some plan of action. Similarly, just as the driving examiner makes a log and report, a good tester will objectively document defects found and the outcome of the test.
So, test activities exist before and after test execution, and we explain those activities in this book. As a tester or test manager, you will be involved in planning and control of the testing, choosing test conditions, designing test cases based on those test conditions, executing them and checking results, evaluating whether enough testing has been done by Examining completion (or exit) criteria, reporting on the testing process and system under test, and presenting test completion (or summary) reports.
1.2.4 When can we meet our test objectives?
Testing Principle - Early testing
Testing activities should start as early as possible in the software or system development life cycle and should be focused on defined objectives.
We can use both dynamic testing and static testing as a means for achieving similar test objectives. Both provide information to improve both the system to be tested, and the development and testing processes. We mentioned above that testing can have different goals and objectives, which often include:
• finding defects;
• gaining confidence in and providing information about the level of quality;
• preventing defects.
Many types of review and testing activities-take place at different stages in the life cycle, as we'll see in Chapter 2. These have different objectives. Early testing - such as early test design and review activities - finds defects early on when they are cheap to find and fix. Once the code is written, programmers and testers often run a set of tests so that they can identify and fix defects in the software. In this 'development testing' (which includes component, inte-gration and system testing), the main objective may be to cause as many fail-ures as possible so that defects in the software are identified and can be fixed. Following that testing, the users of the software may carry out acceptance testing to confirm that the system works as expected and to gain confidence that it has met the requirements.
Fixing the defects may not always be the test objective or the desired outcome. Sometimes we simply want to gather information and measure the software. This can take the form of attribute measures such as mean time between failures to assess reliability, or an assessment of the defect density in the software to assess and understand the risk of releasing it.
When maintaining software by enhancing it or fixing bugs, we are changing software that is already being used. In that case an objective of testing may be to ensure that we have not made errors and introduced defects when we changed the software. This is called regression testing - testing to ensure nothing has changed that should not have changed.
We may continue to test the system once it is in operational use. In this case, the main objective may be to assess system characteristics such as reliability or availability.
Testing Principle - Defect clustering
A small number of modules contain most of the defects discovered during pre-release testing or show the most operational failures.
1.2.5 Focusing on defects can help us plan our tests
Reviewing defects and failures in order to improve processes allows us to improve our testing and our requirements, design and development processes. One phenomenon that many testers have observed is that defects tend to cluster. This can happen because an area of the code is particularly complex and tricky, or because changing software and other products tends to cause knock-on defects. Testers will often use this information when making their risk assessment for planning the tests, and will focus on known 'hot spots'.
A main focus of reviews and other static tests is to carry out testing as early as possible, finding and fixing defects more cheaply and preventing defects from appearing at later stages of this project. These activities help us find out about defects earlier and identify potential clusters. Additionally, an important outcome of all testing is information that assists in risk assess-ment; these reviews will contribute to the planning for the tests executed later in the software development life cycle. We might also carry out root cause analysis to prevent defects and failures happening again and perhaps to identify the cause of clusters and potential future clusters.
1.2.6 The defect clusters change over time
Testing Principle - Pesticide paradox
If the same tests are repeated over and over again, eventually the same set of test cases will no longer find any new bugs. To overcome this 'pesticide paradox', the test cases need to be regularly reviewed and revised, and new and different tests need to be written to exercise different parts of the software or system to potentially find more defects.
Over time, as we improve our whole software development life cycle and the early static testing, we may well find that dynamic test levels find fewer defects. A typical test improvement initiative will initially find more defects as the testing improves and then, as the defect prevention kicks in, we see the defect numbers dropping, as shown in Figure 1.3. The first part of the improvement enables us to reduce failures in operation; later the improve-ments are making us more efficient and effective in producing the software with fewer defects in it.
As the 'hot spots' for bugs get cleaned up we need to move our focus else-where, to the next set of risks. Over time, our focus may change from finding coding bugs, to looking at the requirements and design documents for defects, and to looking for process improvements so that we prevent defects in the product. Referring to Figure 1.3, by releases 9 and 10, we would expect that the overall cost and effort associated with reviews and testing is much lower than in releases 4 or 7.
1.2.7 Debugging removes defects
When a test finds a defect that must be fixed, a programmer must do some work to locate the defect in the code and make the fix. In this process, called debug-ging, a programmer will examine the code for the immediate cause of the problem, repair the code and check that the code now executes as expected. The fix is often then tested separately (e.g. by an independent tester) to confirm the fix. Notice that testing and debugging are different activities. Developers may test their own fixes, in which case the very tight cycle of identifying faults,
debugging, and retesting is often loosely referred to as debugging. However, often following the debugging cycle the fixed code is tested independently both to retest the fix itself and to apply regression testing to the surrounding unchanged software.
1.2.8 Is the software defect free?
Testing Principle - Testing shows presence of defects
Testing can show that defects are present, but cannot prove that there are no defects. Testing reduces the probability of undiscovered defects remaining in the software but, even if no defects are found, it is not a proof of correctness.
This principle arises from the theory of the process of scientific experimenta-tion and has been adopted by testers; you'll see the idea in many testing books. While you are not expected to read the scientific theory [Popper] the analogy used in science is useful; however many white swans we see, we cannot say 'All swans are white'. However, as soon as we see one black swan we can say 'Not all swans are white'. In the same way, however many tests we execute without finding a bug, we have not shown 'There are no bugs'. As soon as we find a bug, we have shown 'This code is not bug-free'.
1.2.9 If we don't find defects does that mean the users will accept the software?
Testing Principle - Absence of errors fallacy
Finding and fixing defects does not help if the system built is unusable and does not fulfill the users' needs and expectations.
There is another important principle we must consider; the customers for soft-ware - the people and organizations who buy and use it to aid in their day-to-day tasks - are not interested in defects or numbers of defects, except when they are directly affected by the instability of the software. The people using soft-ware are more interested in the software supporting them in completing tasks efficiently and effectively. The software must meet their needs. It is for this reason that the requirements and design defects we discussed earlier are so important, and why reviews and inspections (see Chapter 3) are such a funda-mental part of the entire test activity.
1.3 TESTING PRINCIPLES
1 Explain the fundamental principles in testing. (K2)
In Sections 1.1 and 1.2, we have introduced a number of testing principles and brief explanations. These are listed in Table 1.2, for you to read over to remind yourself about them. These principles have been suggested over the past 40 years and offer general guidelines common for all testing.
TABLE 1.2 Testing principles
Principle 1: Testing shows Testing can show that defects are present,
presence of defects but cannot prove that there are no
defects. Testing reduces the probability of undiscovered defects remaining in the software but, even if no defects are found, it is not a proof of correctness.
Principle 2: Exhaustive testing Testing everything (all combinations of is impossible inputs and preconditions) is not feasible
except for trivial cases. Instead of exhaustive testing, we use risks and priorities to focus testing efforts.
Principle 3: Early testing Testing activities should start as early as
possible in the software or system development life cycle and should be focused on defined objectives.
Principle 4: Defect clustering A small number of modules contain most
of the defects discovered during pre-release testing or show the most operational failures.
Principle 5: Pesticide paradox If the same tests are repeated over and
over again, eventually the same set of test cases will no longer find any new bugs. To overcome this 'pesticide paradox', the test cases need to be regularly reviewed and revised, and new and different tests need to be written to exercise different parts of the software or system to potentially find more defects.
Principle 6: Testing is context Testing is done differently in different dependent contexts. For example, safety-critical
software is tested differently from an e-commerce site.
Principle 7: Absence-of-errors Finding and fixing defects does not help if fallacy the system built is unusable and does not
fulfill the users' needs and expectations.
1.4 FUNDAMENTAL TEST PROCESS
1 Recall the fundamental test activities from planning to test closure activities and the main tasks of each test activity. (Kl)
1.4.1 Introduction
In this section, we will describe the fundamental test process and activi-ties. These start with test planning and continue through to test closure. For each part of the test process, we'll discuss the main tasks of each test activity.
In this section, you'll also encounter the glossary terms confirmation testing, exit criteria, incident, regression testing, test basis, test condition, test coverage, test data, test execution, test log, test plan, test strategy, test summary report and testware.
As we have seen, although executing tests is important, we also need a plan of action and a report on the outcome of testing. Project and test plans should include time to be spent on planning the tests, designing test cases, preparing for execution and evaluating status. The idea of a fundamental test process for all levels of test has developed over the years. Whatever the level of testing, we see the same type of main activities happening, although there may be a different amount of formality at the different levels, for example, component tests might be carried out less formally than system tests in most organizations with a less documented test process. The decision about the level of formality of the processes will depend on the system and software context and the level of risk associated with the software. So we can divide the activities within the fundamental test process into the following basic steps:
• planning and control;
• analysis and design;
• implementation and execution;
• evaluating exit criteria and reporting;
• test closure activities.
These activities are logically sequential, but, in a particular project, may overlap, take place concurrently and even be repeated. This process is par-ticularly used for dynamic testing, but the main headings of the process can be applied to reviews as well. For example, we need to plan and prepare for reviews, carry out the reviews, and evaluate the outcomes of the reviews. For some reviews, such as inspections, we will have exit crite-ria and will go through closure activities. However, the detail and naming of the activities will be different for static testing. We'll discuss static testing in Chapter 3.
1.4.2 Test planning and control
During test planning, we make sure we understand the goals and objectives of the customers, stakeholders, and the project, and the risks which testing is intended to address. This will give us what is sometimes called the mission of testing or the test assignment. Based on this understanding, we set the goals and objectives for the testing itself, and derive an approach and plan for the tests, including specification of test activities. To help us we may have organization or program test policies and a test strategy. Test policy gives rules for testing, e.g. 'we always review the design documents'; test strategy is the overall high-level approach, e.g. 'system testing is carried out by an independent team reporting to the program quality manager. It will be risk-based and proceeds from a product (quality) risk analysis' (see Chapter 5). If policy and strategy are defined already they drive our planning but if not we should ask for them to be stated and defined. Test planning has the following major tasks, given approxi-mately in order, which help us build a test plan:
• Determine the scope and risks and identify the objectives of testing: we con sider what software, components, systems or other products are in scope for testing; the business, product, project and technical risks which need to be addressed; and whether we are testing primarily to uncover defects, to show that the software meets requirements, to demonstrate that the system is fit for purpose or to measure the qualities and attributes of the software.
• Determine the test approach (techniques, test items, coverage, identifying and interfacing with the teams involved in testing, testware): we consider how we will carry out the testing, the techniques to use, what needs testing and how extensively (i.e. what extent of coverage). We'll look at who needs to get involved and when (this could include developers, users, IT infrastruc ture teams); we'll decide what we are going to produce as part of the testing (e.g. testware such as test procedures and test data). This will be related to the requirements of the test strategy.
• Implement the test policy and/or the test strategy: we mentioned that there may be an organization or program policy and strategy for testing. If this is the case, during our planning we must ensure that what we plan to do adheres to the policy and strategy or we must have agreed with stakeholders, and documented, a good reason for diverging from it.
• Determine the required test resources (e.g. people, test environment, PCs): from the planning we have already done we can now go into detail; we decide on our team make-up and we also set up all the supporting hardware and software we require for the test environment.
• Schedule test analysis and design tasks, test implementation, execution and evaluation: we will need a schedule of all the tasks and activities, so that we can track them and make sure we can complete the testing on time.
• Determine the exit criteria: we need to set criteria such as coverage criteria (for example, the percentage of statements in the software that must be executed during testing) that will help us track whether we are completing the test activ ities correctly. They will show us which tasks and checks we must complete for a particular level of testing before we can say that testing is finished.
Management of any activity does not stop with planning it. We need to control and measure progress against the plan. So, test control is an ongoing activity. We need to compare actual progress against the planned progress, and report to the project manager and customer on the current status of testing, including any changes or deviations from the plan. We'll need to take actions where necessary to meet the objectives of the project. Such actions may entail changing our original plan, which often happens. When different groups perform different review and test activities within the project, the planning and control needs to happen within each of those groups but also across the groups to coordinate between them, allowing smooth hand-offs between each stage of testing. Test planning takes into account the feedback from monitoring and control activities which take place through out the project. Test control has the following major tasks:
• Measure and analyze the results of reviews and testing: We need to know how many reviews and tests we have done. We need to track how many tests have passed and how many failed, along with the number, type and importance of the defects reported.
• Monitor and document progress, test coverage and exit criteria: It is important that we inform the project team how much testing has been done, what the results are, and what conclusions and risk assessment we have made. We must make the test outcome visible and useful to the whole team.
• Provide information on testing: We should expect to make regular and exceptional reports to the project manager, project sponsor, customer and other key stakeholders to help them make informed decisions about project status. We should also use the information we have to analyze the testing itself.
• Initiate corrective actions: For example, tighten exit criteria for defects fixed, ask for more effort to be put into debugging or prioritize defects for fixing test blockers.
• Make decisions: Based on the measures and information gathered during testing and any changes to business and project risks or our increased under standing of technical and product risks, we'll make decisions or enable others to make decisions: to continue testing, to stop testing, to release the software or to retain it for further work for example.
1.4.3 Test analysis and design
Test analysis and design is the activity where general testing objectives are trans-formed into tangible test conditions and test designs. During test analysis and design, we take general testing objectives identified during planning and build test designs and test procedures (scripts). You'll see how to do this in Chapter 4. Test analysis and design has the following major tasks, in approximately the following order:
• Review the test basis (such as the product risk analysis, requirements, archi tecture, design specifications, and interfaces), examining the specifications for the software we are testing. We use the test basis to help us build our tests. We can start designing certain kinds of tests (called black-box tests)
before the code exists, as we can use the test basis documents to understand what the system should do once built. As we study the test basis, we often identify gaps and ambiguities in the specifications, because we are trying to identify precisely what happens at each point in the system, and this also pre-vents defects appearing in the code.
• Identify test conditions based on analysis of test items, their specifications, and what we know about their behavior and structure. This gives us a high- level list of what we are interested in testing. If we return to our driving example, the examiner might have a list of test conditions including 'behav ior at road junctions', 'use of indicators', 'ability to maneuver the car' and so on. In testing, we use the test techniques to help us define the test condi tions. From this we can start to identify the type of generic test data we might need.
• Design the tests (you'll see how to do this in Chapter 4), using techniques to help select representative tests that relate to particular aspects of the soft ware which carry risks or which are of particular interest, based on the test conditions and going into more detail. For example, the driving examiner might look at a list of test conditions and decide that junctions need to include T-junctions, cross roads and so on. In testing, we'll define the test case and test procedures.
• Evaluate testability of the requirements and system. The requirements may be written in a way that allows a tester to design tests; for example, if the per formance of the software is important, that should be specified in a testable way. If the requirements just say 'the software needs to respond quickly enough' that is not testable, because 'quick enough' may mean different things to different people. A more testable requirement would be 'the soft ware needs to respond in 5 seconds with 20 people logged on'. The testabil ity of the system depends on aspects such as whether it is possible to set up the system in an environment that matches the operational environment and whether all the ways the system can be configured or used can be understood and tested. For example, if we test a website, it may not be possible to iden tify and recreate all the configurations of hardware, operating system, browser, connection, firewall and other factors that the website might encounter.
• Design the test environment set-up and identify any required infrastructure and tools. This includes testing tools (see Chapter 6) and support tools such as spreadsheets, word processors, project planning tools, and non-IT tools and equipment - everything we need to carry out our work.
1.4.4 Test implementation and execution
During test implementation and execution, we take the test conditions and make them into test cases and testware and set up the test environment. This means that, having put together a high-level design for our tests, we now start to build them. We transform our test conditions into test cases and procedures, other testware such as scripts for automation. We also need to set up an envi-ronment where we will run the tests and build our test data. Setting up environ-ments and data often involves significant time and effort, so you should plan
and monitor this work carefully. Test implementation and execution have the following major tasks, in approximately the following order:
• Implementation:
- Develop and prioritize our test cases, using the techniques you'll see in Chapter 4, and create test data for those tests. We will also write instructions for carrying out the tests (test procedures). For the driving examiner this might mean changing the test condition 'junc tions' to 'take the route down Mayfield Road to the junction with Summer Road and ask the driver to turn left into Summer Road and then right into Green Road, expecting that the driver checks mirrors, signals and maneuvers correctly, while remaining aware of other road users.' We may need to automate some tests using test harnesses and automated test scripts. We'll talk about automation more in Chapter 6.
- Create test suites from the test cases for efficient test execution. A test suite is a logical collection of test cases which naturally work together. Test suites often share data and a common high-level set of objectives. We'll also set up a test execution schedule.
- Implement and verify the environment. We make sure the test envi ronment has been set up correctly, possibly even running specific tests on it.
• Execution:
- Execute the test suites and individual test cases, following our test proce dures. We might do this manually or by using test execution tools, accord ing to the planned sequence.
- Log the outcome of test execution and record the identities and versions of the software under test, test tools and testware. We must know exactly what tests we used against what version of the software; we must report defects against specific versions; and the test log we keep provides an audit trail.
- Compare actual results (what happened when we ran the tests) with expected results (what we anticipated would happen).
- Where there are differences between actual and expected results, report discrepancies as incidents. We analyze them to gather further details about the defect, reporting additional information on the problem, identify the causes of the defect, and differentiate between problems in the software and other products under test and any defects in test data, in test documents, or mistakes in the way we exe cuted the test. We would want to log the latter in order to improve the testing itself.
- Repeat test activities as a result of action taken for each discrepancy. We need to re-execute tests that previously failed in order to confirm a fix (confirmation testing or re-testing). We execute corrected tests and suites if there were defects in our tests. We test corrected software again to ensure that the defect was indeed fixed correctly (confirmation test) and that the programmers did not introduce defects in unchanged areas of the software and that fixing a defect did not uncover other defects (regression testing).
1.4.5 Evaluating exit criteria and reporting
Evaluating exit criteria is the activity where test execution is assessed against the defined objectives. This should be done for each test level, as for each we need to know whether we have done enough testing. Based on our risk assess-ment, we'll have set criteria against which we'll measure 'enough'. These crite-ria vary for each project and are known as exit criteria. They tell us whether we can declare a given testing activity or level complete. We may have a mix of cov-erage or completion criteria (which tell us about test cases that must be included, e.g. 'the driving test must include an emergency stop' or 'the software test must include a response measurement'), acceptance criteria (which tell us how we know whether the software has passed or failed overall, e.g. 'only pass the driver if they have completed the emergency stop correctly' or 'only pass the software for release if it meets the priority 1 requirements list') and process exit criteria (which tell us whether we have completed all the tasks we need to do, e.g. 'the examiner/tester has not finished until they have written and filed the end of test report'). Exit criteria should be set and evaluated for each test level. Evaluating exit criteria has the following major tasks:
• Check test logs against the exit criteria specified in test planning: We look to see what evidence we have for which tests have been executed and checked, and what defects have been raised, fixed, confirmation tested, or are out standing.
• Assess if more tests are needed or if the exit criteria specified should be changed: We may need to run more tests if we have not run all the tests we designed, or if we realize we have not reached the coverage we expected, or if the risks have increased for the project. We may need to change the exit criteria to lower them, if the business and project risks rise in impor tance and the product or technical risks drop in importance. Note that this is not easy to do and must be agreed with stakeholders. The test manage ment tools and test coverage tools that we'll discuss in Chapter 6 help us with this assessment.
• Write a test summary report for stakeholders: It is not enough that the testers know the outcome of the test. All the stakeholders need to know what testing has been done and the outcome of the testing, in order to make informed decisions about the software.
1.4.6 Test closure activities
During test closure activities, we collect data from completed test activities to consolidate experience, including checking and filing testware, and analyzing facts and numbers. We may need to do this when software is delivered. We also might close testing for other reasons, such as when we have gathered the infor-mation needed from testing, when the project is cancelled, when a particular milestone is achieved, or when a maintenance release or update is done. Test closure activities include the following major tasks:
• Check which planned deliverables we actually delivered and ensure all incident reports have been resolved through defect repair or deferral. For deferred defects, in other words those that remain open, we may request
a change in a future release. We document the-acceptance or rejection of the software system.
• Finalize and archive testware, such as scripts, the test environment, and any other test infrastructure, for later reuse. It is important to reuse whatever we can of testware; we will inevitable carry out maintenance testing, and it saves time and effort if our testware can be pulled out from a library of existing tests. It also allows us to compare the results of testing between software versions.
• Hand over testware to the maintenance organization who will support the software and make any bug fixes or maintenance changes, for use in con firmation testing and regression testing. This group may be a separate group to the people who build and test the software; the maintenance testers are one of the customers of the development testers; they will use the library of tests.
• Evaluate how the testing went and analyze lessons learned for future releases and projects. This might include process improvements for the soft ware development life cycle as a whole and also improvement of the test processes. If you reflect on Figure 1.3 again, we might use the test results to set targets for improving reviews and testing with a goal of reducing the number of defects in live use. We might look at the number of incidents which were test problems, with the goal of improving the way we design, execute and check our tests or the management of the test environments and data. This helps us make our testing more mature and cost-effective for the organization. This is documented in a test summary report or might be part of an overall project evaluation report.
1.5 THE PSYCHOLOGY OF TESTING
1 Recall that the success of testing is influenced by psychological factors:
(Kl)
• clear objectives;
• a balance of self-testing and independent testing;
• recognition of courteous communication and feedback on defects.
2 Contrast the mindset of a tester and that of a developer. (K2)
In this section, we'll discuss the various psychological factors that influence testing and its success. These include clear objectives for testing, the proper roles and balance of self-testing and independent testing, clear, courteous com-munication and feedback on defects. We'll also contrast the mindset of a tester and of a developer.
You'll find a single Syllabus term in this section, independent testing, and the glossary term, independence.
1.5.1 Independent testing - who is a tester?
The mindset we want to use while testing and reviewing is different from the one we use while analyzing or developing. By this we mean that, if we are build-ing something we are working positively to solve problems in the design and to realize a product that meets some need. However, when we test or review a product, we are looking for defects in the product and thus are critical of it.
Suppose you were going to cook a meal to enter in a competition for chefs. You select the menu, collect the ingredients, cook the food, set the table, and serve the meal. If you want to win, you do each task as well as you can. Suppose instead you are one of the judges evaluating the competition meals. You examine everything critically, including the menu, the ingredients, the methods used, keeping to time and budget allowances, choice of ingredients, the ele-gance of the table setting and the serving, and the look and taste of the meal. To differentiate between the competition chefs, you'll praise every good aspect of their performances but you'll also note every fault and error each chef made. So it is with software testing: building the software requires a different mindset from testing the software.
We do not mean that a tester cannot be a programmer, or that a program-mer cannot be a tester, although they often are separate roles. In fact, program-mers are testers - they test the components which they build, and the integration of the components into the system. The good chef will be as critical as the competition judges of his own work, in order to prevent and rectify errors and defects before anyone notices them. So, with the right mindset, program-mers can test their own code; indeed programmers do test their own code and find many problems, resolving them before anyone else sees the code. Business analysis and marketing staff should review their own requirements. System architects should review their own designs. However, we all know it is difficult to find our own mistakes. So, business analysts, marketing staff, architects and programmers often rely on others to help test their work. This other person might be a fellow analyst, designer or developer. A person who will use the soft-ware may help test it. Business analysts who worked on the requirements and design may perform some tests. Testing specialists - professional testers - are often involved. In fact, testing may involve a succession of people each carrying out a different level of testing. This allows an independent test of the system.
We'll look at the points in the software development life cycle where testing takes place in Chapter 2. You'll see there that several stages of reviews and testing are carried out throughout the life cycle and these may be independent reviews and tests. Early in the life cycle, reviews of requirements and design documents by someone other than the author helps find defects before coding starts and helps us build the right software. Following coding, the software can be tested independently. This degree of independence avoids author bias and is often more effective at finding defects and failures.
Several levels of independence can be identified, listed here from the lowest level of independence to the highest:
• tests by the person who wrote the item under test;
• tests by another person within the same team, such as another programmer;
• tests by a person from a different organizational group, such as an independ ent test team;
• tests designed by a person from a different-organization or company, such as outsourced testing or certification by an external body.
We should note, however, that independence is not necessarily the most important factor in good testing. Developers who know how to test and who are, like good chefs, self-critical, have the benefit of familiarity and the pride-of-work that comes with true professionalism. Such developers can efficiently find many defects in their own code. Some software development methodolo-gies insist on developers designing tests before they start coding and executing those tests continuously as they change the code. This approach promotes early testing and early defect detection, which is cost effective. Remember, independ-ent testing may be carried out at any level of testing and the choice of independ-ence level depends on the risk in the particular context.
1.5.2 Why do we sometimes not get on with the rest of the team?
As well as independence, separation of the tester role from the developer role is also done to help focus effort and to provide the benefits of trained and professional testing resources. In many organizations, earlier stages of testing are carried out by the developers and integrators and later stages independently, either by a specialist test group or by the customers. However, this separation can lead to problems as well as advantages. The advantage of independence and focus may be lost if the inter-team relation-ships deteriorate, as we'll see.
Each organization and each project will have its own goals and objectives. Different stakeholders, such as the customers, the development team and the managers of the organization, will have different viewpoints about quality and have their own objectives. Because people and projects are driven by objectives, the stakeholder with the strongest views or the greatest influence over a group will define, consciously or subconsciously, what those objectives are. People tend to align their plans with these objectives. For example, depending on the objective, a tester might focus either on finding defects or on confirming that software works. But if one stakeholder is less influential during the project but more influential at delivery, there may be a clash of views about whether the testing has met its objectives. One manager may want the confirmation that the software works and that it is 'good enough' if this is seen as a way of delivering as fast as possible. Another manager may want the testing to find as many defects as possible before the software is released, which will take longer to do and will require time for fixing, re-testing and regression testing. If there are not clearly stated objectives and exit criteria for testing which all the stakeholders have agreed, arguments might arise, during the testing or after release, about whether 'enough' testing has been done.
Many of us find it challenging to actually enjoy criticism of our work. We usually believe that we have done our best to produce work (documents, code, tests, whatever) which is correct and complete. So when someone else identifies a defect, a mistake we have made, we might take this personally and get annoyed with the other person, especially if we are under time pressure. This is true of managers, staff, testers and developers. We all make mistakes and we sometimes get annoyed, upset or depressed when someone points them out. So,
when as testers we run a test which (from our viewpoint) is a good test that finds defects and failures in the software, we need to be careful how we react. We are pleased, of course, since we have found a good bug! But how will the require-ments analyst, designer, developer, project manager and customer react? The people who build products may react defensively and perceive this reported defect as personal criticism against the product and against the author. The project manager may be annoyed with everyone for holding up the project. The customer may lose confidence in the product because he can see defects. Because testing can be seen as a destructive activity, we need to take care to report on defects and failures as objectively and politely as possible. If others are to see our work as constructive in the management of product risks, we need to be careful when we are reviewing and when we are testing:
• Communicate findings on the product in a neutral, fact-focused way without criticizing the person who created it. For example, write objective and factual incident reports and review findings.
- Don't gloat - you are not perfect either!
- Don't blame - any mistakes are probably by the group rather than an individual.
- Be constructively critical and discuss the defect and how you are going to log it.
• Explain that by knowing about this now we can work round it or fix it so the delivered system is better for the customer.
- Say what you liked and what worked, as well as what didn't work.
- Show what the risk is honestly - not everything is high priority.
- Don't just see the pessimistic side - give praise as well as criticism.
- Show what risks have been uncovered and the benefits of the review or test.
• Start with collaboration rather than battles. Remind everyone of the common goal of better quality systems.
- Be polite and helpful, collaborate with your colleagues.
- Try to understand how the other person feels and why they react as they do.
- Confirm that the other person has understood what you have said and vice versa.
- Explain how the test or review helps the author - what's in it for him or her.
- Offer your work to be reviewed, too.
It's our job as reviewers and testers to provide everyone with clear, objective information and to do this we go bug-hunting, defect-mining and failure-making. People who will make good reviewers and testers have the desire and ability to find problems, and this is true whether testing is their main job or part of their role as a developer. These people build up experience of where errors are likely to be made, and are characterized by their curiosity, professional pes-simism, critical eye and attention to detail. However, unless we also have good interpersonal and communication skills, courtesy, understanding of others and a good attitude towards our peers, colleagues, customers, managers and the rest of the team, we will fail as testers because no-one will listen to us.
The tester and test leader need good interpersonal skills to communicate factual information about defects, progress and risks in a constructive way [Perry]. For the author of the software or document, defect information can help them improve their skills, but only if it is provided in a way that helps them. One book that you might find interesting in this context is Six Thinking Hats [de Bono]. It is not about testing but describes a way to communicate different information: facts; our emotions; pessimistic and optimistic thoughts; and cre-ative ideas. When reviewing or testing, we need to communicate facts objec-tively, but the other types of information are useful too: 'This happened; this is how I felt about it; this is what was good; this is what might go wrong; here is a way we could solve the problem'. As part of supplying the risk assessment, we can help the managers and customers make risk-based decisions based on the cost and time impact of a defect. If we test and find a defect that would cost $15 000 to fix and re-test/regression test, is it worth fixing? If it would cause a business impact of $50 000 in the live environment the customer may want it fixed. If it has a potential business impact of $10 000 but any fix is difficult to do and likely to have adverse impact elsewhere, it may be better not to fix. By pro-viding the team with information about the defect in terms they find useful, we can help them to make the right decision about fixing or not fixing the prob-lems. Generally we say that defects found and fixed during testing will save time and money later and reduce risks, so we need to show that is the case in order for the testing to be valued.
To help you think about the psychology of testing, there is an exercise at the end of the chapter, following the practice examination questions.
CHAPTER REVIEW
Let's review what you have learned in this chapter.
From Section 1.1, you should now be able to explain why testing is necessary and support that explanation with examples and evidence. You should be able to give examples of negative consequences of a software defect or bug for people, companies, and the environment. You should be able to contrast a defect with its symptoms. You should be able to discuss the ways in which testing fits into and supports higher quality. You should know the glossary terms bug, defect, error, failure, fault, mistake, quality, risk, software, testing and exhaustive testing.
From Section 1.2, you should now know what testing is. You should be able to remember the common objectives of testing. You should be able to describe how testing can find defects, provide confidence and information and prevent defects. You should be able to explain the fundamental principles of testing, summarized in Section 1.3. You should know the glossary terms code, debug-ging, development of software, requirement, review, test basis, test case, testing and test objective.
From Section 1.4, you should now recognize the fundamental test process, as well as being aware of some other related ways to model the test process. You should be able to recall the main testing activities related to test planning and control, analysis and design, implementation and execution, evaluating exit cri-teria and reporting, and test closure. You should know the glossary terms con-firmation testing, exit criteria, incident, regression testing, test basis, test condition, test coverage, test data, test execution, test log, test plan, test strategy, test summary report and testware.
Finally, from Section 1.5, you now should be able to explain the psychology of testing and how people influence testing success. You should recall the importance of clear objectives, the right mix of self-testing and independent testing and courteous, respectful communication between testers and others on the project team, especially about defects. You should be able to explain and contrast the mindsets of testers and programmers and why these differences can lead to conflicts. You should know the glossary term independence.
SAMPLE EXAM QUESTIONS
Question 1 A company recently purchased a commercial off-the-shelf application to automate their bill-paying process. They now plan to run an acceptance test against the package prior to putting it into production. Which of the following is their most likely reason for testing?
a. To build confidence in the application.
b. To detect bugs in the application.
c. To gather evidence for a lawsuit.
d. To train the users.
Question 2 According to the ISTQB Glossary, the word 'bug' is synonymous with which of the following words?
a. Incident
b. Defect
c. Mistake
d. Error
Question 3 According to the ISTQB Glossary, a risk relates to which of the following?
a. Negative feedback to the tester.
b. Negative consequences that will occur.
c. Negative consequences that could occur.
d. Negative consequences for the test object.
Question 4 Ensuring that test design starts during the requirements definition phase is impor-tant to enable which of the following test objec-tives?
a. Preventing defects in the system.
b. Finding defects through dynamic testing.
c. Gaining confidence in the system.
d. Finishing the project on time.
Question 5 A test team consistently finds between 90% and 95% of the defects present in the system under test. While the test manager understands that this is a good defect-detection percentage for her test team and industry, senior management and executives remain disappointed in the test group, saying that the test team misses too many bugs. Given that the users are generally
happy with the system and that the failures which have occurred have generally been low impact, which of the following testing principles is most likely to help the test manager explain to these managers and executives why some defects are likely to be missed?
a. Exhaustive testing is impossible
b. Defect clustering
c. Pesticide paradox
d. Absence-of-errors fallacy
Question 6 According to the ISTQB Glossary, regression testing is required for what purpose?
a. To verify the success of corrective actions.
b. To prevent a task from being incorrectly consid ered completed.
c. To ensure that defects have not been introduced by a modification.
d. To motivate better unit testing by the program mers.
Question 7 Which of the following is most important to promote and maintain good relation-ships between testers and developers?
a. Understanding what managers value about testing.
b. Explaining test results in a neutral fashion.
c. Identifying potential customer work-arounds for bugs.
d. Promoting better quality software whenever possible.
Question 8 Which of the statements below is the best assessment of how the test principles apply across the test life cycle?
a. Test principles only affect the preparation for testing.
b. Test principles only affect test execution activi ties.
c. Test principles affect the early test activities such as review.
d. Test principles affect activities throughout the test life cycle.
EXERCISE: TEST PSYCHOLOGY
Read the email below, and see what clues you find to help you identify problems in the scenario described. Categorize the clues/problems as:
• possible people, psychology and attitude problems;
• other problems, e.g. possible test management and role problems, possible product problems.
Hi there!
Well, I nearly caused a panic today because I thought I had found a mega showstopper on the trading system we are testing. The test manager and others got involved examining databases first on the server and then on the gateway that feeds the clients, checking update logs from processes that ran overnight as well as checking data passed to the client. Eventually I found the problem. I had mis-clicked on a .bat file when running up a client and had run up the wrong client environment. By that time the test manager was ready to say a few short words in my ear, particularly as the development people had started to get involved and they have zero tolerance for mistakes made by testers. The only saving grace was that I found the mistake and not one of the developers.
It was, objectively, an interesting mistake. When you log into the server test environments, the panels always show the environment to which you are connected. In our case we have two test environments called Systest14 and Systest15 and my tests were set up in Systest15. To run up the clients, we have to run .bat files for either a 14 or 15 client. I had started two clients, that is two exchange participants, so I could do some trading between them.
It appears I started the first client OK in environment 15 but when I started the second, I accidentally moved the mouse a fraction so it ran the 14 .bat file that is next to it in the Explorer file list. To make matters worse, the client screens do not show the environment to which you are attached.
At first I felt a bit stupid having caused much hectic and wasted activity. On reflection I thought that if I, as a reasonably competent person, can make a mistake like this then something is wrong. On the server side when I log on to a test environment, I have to enter the environment name and it's shown on all the panels. On the client side, I run a client test environment by selecting a .bat file from a list of many and have to ensure I click on the right file. There is neither a display nor the ability to determine the client environment in which I am working.
So I am going to log this as a high priority, or even showstopper, error - the client does not show the environment. In real life terms, it means a real user could be connected to the production system and think he is connected to a test system and screw up trading. I know this happened once on the equities trading system, when a trader entered a load of test transactions into the production system by mistake and caused mayhem.
As an addendum to this story, a couple of days later one of the testers found what appeared to be another mega showstopper. He and the test manager spent three hours crawling all over the system before they discovered the 'error'. A new filter had been added to the client software to filter transactions displayed in panels by geographical market. Unknown to them, it was set to a default of the German market, whereas they thought they were in the UK market. Consequently, at first sight, it appeared there were fundamental problems with the network transaction bus and the message-broadcasting systems. Apart from the issue that they should have been informed of this change, it raised a similar problem to the one I had experienced -the client system does not display the market in which you are trading.
Well - I'm off for another happy day at the office! All the best
EXERCISE SOLUTION
People, psychology and attitude problems include, for example:
• Poor relationships between the test team and the test manager, and the testers and developers, e.g. 'By that time the test manager was ready to say a few short words in my ear, particularly as the development people had started to get involved and they have zero tolerance for mistakes made by testers. The only saving grace was that I found the mistake and not one of the developers.'
• Emotive use of language - understandable in the circumstances but not suitable for reporting problems, e.g. 'Well, I nearly caused a panic today because I thought I had found a mega showstopper on the trading system we are testing,' and 'As an addendum to this story, a couple of days later one of the testers found what appeared to be another mega-showstopper.'
• Initial diffidence overcome by revisiting the problem - if one person can make this mistake then others will. 'At first I felt a bit stupid having caused much hectic and wasted activity. On reflection I thought that if I, as a reasonably competent person, can make a mistake like this then something is wrong.'
• Understandable use of sarcasm ... 'Well - I'm off for another happy day at the office!' Other problems include test management and role problems, for example:
• Configuration management and release control - A new filter had been added to the client software to filter transactions displayed in panels by geographical market.'
• Configuration management, relationships, communications - Apart from the issue that they should have been informed of this change ....'
• Does the test manager really understand his role? 'He and the test manager spent three hours crawling all over the system before they discovered the "error",' and 'The test manager and others got involved examining databases.'
There are some product problems, although no functionality or technical problems. Not all the problems we encounter as testers are functionality or technical problems. There are some non-functional problems - specifically, usability - which indicate that a real user might be inconvenienced or worse by this problem:
• 'I had mis-clicked on a .bat file ...'
• 'In real life terms, it means a real user could be connected to the production system and think he is connected to a test system and screw up trading. I know this happened once ... when a trader entered a load of test transactions into the production system by mistake and caused mayhem.'
• 'It raised a similar problem to the one I had experienced - the client system does not display the market in which you are trading.'
• 'There is neither a display nor the ability to determine the client environment in which I am working.' And 'To make matters worse, the client screens do not show the environment to which you are attached.'
• 'Unknown to them, it was set to a default of the German market, whereas they thought they were in the UK market.'
Note that we will return to this exercise at the end of Chapter 5, where we deal with writing a good incident report.
CHAPTER 2 Testing throughout the software life cycle
esting is not a stand-alone activity. It has its place within a software development life cycle model and therefore the life cycle applied will largely determine how testing is organized. There are many different forms of testing. Because several disciplines, often with different interests, are involved in the development life cycle, it is important to clearly understand and define the various test levels and types. This chapter discusses the most commonly applied software development models, test levels and test types. Maintenance can be seen as a specific instance of a development process. The way maintenance influences the test process, levels and types and how testing can be organized is described in the last section of this chapter.
2.1 SOFTWARE DEVELOPMENT MODELS
1 Understand the relationship between development, test activities and work products in the development life cycle and give examples based on project and product characteristics and context. (K2)
2 Recognize the fact that software development models must be adapted to the context of project and product characteristics. (Kl)
3 Recall reasons for different levels of testing and characteristics of good testing in any life cycle model. (Kl)
The development process adopted for a project will depend on the project aims and goals. There are numerous development life cycles that have been developed in order to achieve different required objectives. These life cycles range from lightweight and fast methodologies, where time to market is of the essence, through to fully controlled and documented methodologies where quality and reliability are key drivers. Each of these methodologies has its place in modern software development and the most appropriate development process should be applied to each project. The models specify the various stages of the process and the order in which they are carried out.
The life cycle model that is adopted for a project will have a big impact on the testing that is carried out. Testing does not exist in isolation; test activities
are highly related to software development activities. It will define the what, where, and when of our planned testing, influence regression testing, and largely determine which test techniques to use. The way testing is organized must fit the development life cycle or it will fail to deliver its benefit. If time to market is the key driver, then the testing must be fast and efficient. If a fully documented software development life cycle, with an audit trail of evidence, is required, the testing must be fully documented.
In every development life cycle, a part of testing is focused on verification testing and a part is focused on validation testing. Verification is concerned with evaluating a work product, component or system to determine whether it meets the requirements set. In fact, verification focuses on the question 'Is the deliverable built according to the specification?'. Validation is concerned with evaluating a work product, component or system to determine whether it meets the user needs and requirements. Validation focuses on the question 'Is the deliverable fit for purpose, e.g. does it provide a solution to the problem?'.
2.1.1 V-model
Before discussing the V-model, we will look at the model which came before it. The waterfall model was one of the earliest models to be designed. It has a natural timeline where tasks are executed in a sequential fashion. We start at the top of the waterfall with a feasibility study and flow down through the various project tasks finishing with implementation into the live environment. Design flows through into development, which in turn flows into build, and finally on into test. Testing tends to happen towards the end of the project life cycle so defects are detected close to the live implementation date. With this model it has been difficult to get feedback passed backwards up the waterfall and there are difficulties if we need to carry out numerous iterations for a particular phase.
The V-model was developed to address some of the problems experienced using the traditional waterfall approach. Defects were being found too late in the life cycle, as testing was not involved until the end of the project. Testing also added lead time due to its late involvement. The V-model pro-vides guidance that testing needs to begin as early as possible in the life cycle, which, as we've seen in Chapter 1, is one of the fundamental princi-ples of structured testing. It also shows that testing is not only an execution-based activity. There are a variety of activities that need to be performed before the end of the coding phase. These activities should be carried out in parallel with development activities, and testers need to work with devel-opers and business analysts so they can perform these activities and tasks and produce a set of test deliverables. The work products produced by the developers and business analysts during development are the basis of testing in one or more levels. By starting test design early, defects are often found in the test basis documents. A good practice is to have testers involved even earlier, during the review of the (draft) test basis documents. The V-model is a model that illustrates how testing activities (verification and validation) can be integrated into each phase of the life cycle. Within the V-model, validation testing takes place especially during the early stages, e.g. reviewing the user requirements, and late in the life cycle, e.g. during user acceptance testing.
Although variants of the V-model exist, a common type of V-model uses four test levels. The four test levels used, each with their own objectives, are:
• component testing: searches for defects in and verifies the functioning of software components (e.g. modules, programs, objects, classes etc.) that are separately testable;
• integration testing: tests interfaces between components, interactions to dif ferent parts of a system such as an operating system, file system and hard ware or interfaces between systems;
• system testing: concerned with the behavior of the whole system/product as defined by the scope of a development project or product. The main focus of system testing is verification against specified requirements;
• acceptance testing: validation testing with respect to user needs, require ments, and business processes conducted to determine whether or not to accept the system.
The various test levels are explained and discussed in detail in Section 2.2.
In practice, a V-model may have more, fewer or different levels of devel-opment and testing, depending on the project and the software product. For example, there may be component integration testing after component testing and system integration testing after system testing. Test levels can be combined or reorganized depending on the nature of the project or the system architecture. For the integration of a commercial off-the-shelf (COTS) software product into a system, a purchaser may perform only inte-gration testing at the system level (e.g. integration to the infrastructure and other systems) and at a later stage acceptance testing.
Note that the types of work products mentioned in Figure 2.2 on the left side of the V-model are just an illustration. In practice they come under many dif-ferent names. References for generic work products include the Capability Maturity Model Integration (CMMi) or the 'Software life cycle processes' from ISO/IEC 12207. The CMMi is a framework for process improvement for both system engineering and software engineering. It provides guidance on where to focus and how, in order to increase the level of process maturity [Chrissis et ah, 2004]. ISO/IEC 12207 is an integrated software life cycle process standard that is rapidly becoming more popular.
2.1.2 Iterative life cycles
Not all life cycles are sequential. There are also iterative or incremental life cycles where, instead of one large development time line from beginning to end, we cycle through a number of smaller self-contained life cycle phases for the same project. As with the V-model, there are many variants of iterative life
cycles.
A common feature of iterative approaches is that the delivery is divided into increments or builds with each increment adding new functionality. The initial increment will contain the infrastructure required to support the initial build functionality. The increment produced by an iteration may be tested at several levels as part of its development. Subsequent increments will need testing for the new functionality, regression testing of the existing functionality, and inte-gration testing of both new and existing parts. Regression testing is increasingly important on all iterations after the first one. This means that more testing will be required at each subsequent delivery phase which must be allowed for in the project plans. This life cycle can give early market presence with critical func-tionality, can be simpler to manage because the workload is divided into smaller pieces, and can reduce initial investment although it may cost more in the long run. Also early market presence will mean validation testing is carried out at each increment, thereby giving early feedback on the business value and fitness-for-use of the product.
Examples of iterative or incremental development models are prototyping, Rapid Application Development (RAD), Rational Unified Process (RUP) and agile development. For the purpose of better understanding iterative develop-ment models and the changing role of testing a short explanation of both RAD and agile development is provided.
Rapid Application Development
Rapid Application Development (RAD) is formally a parallel development of
functions and subsequent integration.
Components/functions are developed in parallel as if they were mini proj-ects, the developments are time-boxed, delivered, and then assembled into a working prototype. This can very quickly give the customer something to see and use and to provide feedback regarding the delivery and their requirements. Rapid change and development of the product is possible using this methodol-ogy. However the product specification will need to be developed for the product at some point, and the project will need to be placed under more formal controls prior to going into production. This methodology allows early
validation of technology risks and a rapid response to changing customer requirements.
Dynamic System Development Methodology [DSDM] is a refined RAD process that allows controls to be put in place in order to stop the process from getting out of control. Remember we still need to have the essentials of good development practice in place in order for these methodologies to work. We need to maintain strict configuration management of the rapid changes that we are making in a number of parallel development cycles. From the testing perspective we need to plan this very carefully and update our plans regularly as things will be changing very rapidly (see Chapter 5 for more on test plans).
The RAD development process encourages active customer feedback. The customer gets early visibility of the product, can provide feedback on the design and can decide, based on the existing functionality, whether to proceed with the development, what functionality to include in the next delivery cycle or even to halt the project if it is not delivering the expected value. An early business-focused solution in the market place gives an early return on investment (ROI) and can provide valuable marketing information for the business. Validation with the RAD development process is thus an early and major activity.
Agile development
Extreme Programming (XP) is currently one of the most well-known agile development life cycle models. (See [Agile] for ideas behind this approach.) The methodology claims to be more human friendly than traditional develop-ment methods. Some characteristics of XP are:
• It promotes the generation of business stories to define the functionality.
• It demands an on-site customer for continual feedback and to define and carry out functional acceptance testing.
• It promotes pair programming and shared code ownership amongst the developers.
• It states that component test scripts shall be written before the code is written and that those tests should be automated.
• It states that integration and testing of the code shall happen several times a day.
• It states that we always implement the simplest solution to meet today's problems.
With XP there are numerous iterations each requiring testing. XP develop-ers write every test case they can think of and automate them. Every time a change is made in the code it is component tested and then integrated with the existing code, which is then fully integration-tested using the full set of test cases. This gives continuous integration, by which we mean that changes are incorporated continuously into the software build. At the same time, all test cases must be running at 100% meaning that all the test cases that have been identified and automated are executed and pass. XP is not about doing extreme activities during the development process, it is about doing known vajue-adding activities in an extreme manner.
2.1.3 Testing within a life cycle model
In summary, whichever life cycle model is being used, there are several charac-teristics of good testing:
• for every development activity there is a corresponding testing activity;
• each test level has test objectives specific to that level;
• the analysis and design of tests for a given test level should begin during the corresponding development activity;
• testers should be involved in reviewing documents as soon as drafts are avail able in the development cycle.
2.2 TEST LEVELS
1 Compare the different levels of testing: major objectives, typical objects of testing, typical targets of testing (e.g. functional or structural) and related work products, people who test, types of defects and failures to be identified. (K2)
The V-model for testing was introduced in Section 2.1. This section looks in more detail at the various test levels. The key characteristics for each test level are discussed and defined to be able to more clearly separate the various test levels. A thorough understanding and definition of the various test levels will identify missing areas and prevent overlap and repetition. Sometimes we may wish to introduce deliberate overlap to address specific risks. Understanding whether we want overlaps and removing the gaps will make the test levels more complementary thus leading to more effective and efficient testing.
2.2.1 Component testing
Component testing, also known as unit, module and program testing, searches for defects in, and verifies the functioning of software (e.g. modules, programs, objects, classes, etc.) that are separately testable.
Component testing may be done in isolation from the rest of the system depend-ing on the context of the development life cycle and the system. Most often stubs and drivers are used to replace the missing software and simulate the interface between the software components in a simple manner. A stub is called from the software component to be tested; a driver calls a component to be tested (see Figure 2.5).
Component testing may include testing of functionality and specific non-functional characteristics such as resource-behavior (e.g. memory leaks), per-formance or robustness testing, as well as structural testing (e.g. decision coverage). Test cases are derived from work products such as the software design or the data model.
Typically, component testing occurs with access to the code being tested and with the support of the development environment, such as a unit test frame-work or debugging tool, and in practice usually involves the programmer who wrote the code. Sometimes, depending on the applicable level of risk, compo-nent testing is carried out by a different programmer thereby introducing inde-pendence. Defects are typically fixed as soon as they are found, without formally recording the incidents found.
One approach in component testing, used in Extreme Programming (XP), is to prepare and automate test cases before coding. This is called a test-first approach or test-driven development. This approach is highly iterative and is based on cycles of developing test cases, then building and integrating small pieces of code, and executing the component tests until they pass.
2.2.2 Integration testing
Integration testing tests interfaces between components, interactions to dif-ferent parts of a system such as an operating system, file system and hard-ware or interfaces between systems. Note that integration testing should be differentiated from other integration activities. Integration testing is often carried out by the integrator, but preferably by a specific integration tester or test team.
There may be more than one level of integration testing and it may be carried out on test objects of varying size. For example:
• component integration testing tests the interactions between software com ponents and is done after component testing;
• system integration testing tests the interactions between different systems and may be done after system testing. In this case, the developing organiza tion may control only one side of the interface, so changes may be destabi lizing. Business processes implemented as workflows may involve a series of systems that can even run on different platforms.
The greater the scope of integration, the more difficult it becomes to isolate failures to a specific interface, which may lead to an increased risk. This leads to varying approaches to integration testing. One extreme is that all compo-nents or systems are integrated simultaneously, after which everything is tested as a whole. This is called 'big-bang' integration testing. Big-bang testing has the advantage that everything is finished before integration testing starts. There is no need to simulate (as yet unfinished) parts. The major disadvantage is that in
general it is time-consuming and difficult to trace the cause of failures with this late integration. So big-bang integration may seem like a good idea when plan-ning the project, being optimistic and expecting to find no problems. If one thinks integration testing will find defects, it is a good practice to consider whether time might be saved by breaking the down the integration test process. Another extreme is that all programs are integrated one by one, and a test is carried out after each step (incremental testing). Between these two extremes, there is a range of variants. The incremental approach has the advantage that the defects are found early in a smaller assembly when it is relatively easy to detect the cause. A disadvantage is that it can be time-consuming since stubs and drivers have to be developed and used in the test. Within incremental inte-gration testing a range of possibilities exist, partly depending on the system architecture:
• Top-down: testing takes place from top to bottom, following the control flow or architectural structure (e.g. starting from the GUI or main menu). Components or systems are substituted by stubs.
• Bottom-up: testing takes place from the bottom of the control flow upwards. Components or systems are substituted by drivers.
• Functional incremental: integration and testing takes place on the basis of the functions or functionality, as documented in the functional specification.
The preferred integration sequence and the number of integration steps required depend on the location in the architecture of the high-risk interfaces. The best choice is to start integration with those interfaces that are expected to cause most problems. Doing so prevents major defects at the end of the inte-gration test stage. In order to reduce the risk of late defect discovery, integra-tion should normally be incremental rather than 'big-bang'. Ideally testers should understand the architecture and influence integration planning. If inte-gration tests are planned before components or systems are built, they can be developed in the order required for most efficient testing.
At each stage of integration, testers concentrate solely on the integration itself. For example, if they are integrating component A with component B they are interested in testing the communication between the components, not the functionality of either one. Both functional and structural approaches may be used. Testing of specific non-functional characteristics (e.g. performance) may also be included in integration testing. Integration testing may be carried out by the developers, but can be done by a separate team of specialist integration testers, or by a specialist group of developers/integrators including non-func-tional specialists.
2.2.3 System testing
System testing is concerned with the behavior of the whole system/product as defined by the scope of a development project or product. It may include tests based on risks and/or requirements specification, business processes, use cases, or other high level descriptions of system behavior, interactions with the oper-ating system, and system resources. System testing is most often the final test on behalf of development to verify that the system to be delivered meets the spec-ification and its purpose may be to find as many defects as possible. Most often
it is carried out by specialist testers that form a dedicated, and sometimes inde-pendent, test team within development, reporting to the development manager or project manager. In some organizations system testing is carried out by a third party team or by business analysts. Again the required level of independ-ence is based on the applicable risk level and this will have a high influence on the way system testing is organized.
System testing should investigate both functional and non-functional requirements of the system. Typical non-functional tests include performance and reliability. Testers may also need to deal with incomplete or undocumented requirements. System testing of functional requirements starts by using the most appropriate specification-based (black-box) techniques for the aspect of the system to be tested. For example, a decision table may be created for com-binations of effects described in business rules. Structure-based (white-box) techniques may also be used to assess the thoroughness of testing elements such as menu dialog structure or web page navigation (see Chapter 4 for more on the various types of technique).
System testing requires a controlled test environment with regard to, amongst other things, control of the software versions, testware and the test data (see Chapter 5 for more on configuration management). A system test is executed by the development organization in a (properly controlled) environ-ment. The test environment should correspond to the final target or production environment as much as possible in order to minimize the risk of environment-specific failures not being found by testing.
2.2.4 Acceptance testing
When the development organization has performed its system test and has cor-rected all or most defects, the system will be delivered to the user or customer for acceptance testing. The acceptance test should answer questions such as: 'Can the system be released?', 'What, if any, are the outstanding (business) risks?' and 'Has development met their obligations?'. Acceptance testing is most often the responsibility of the user or customer, although other stakehold-ers may be involved as well. The execution of the acceptance test requires a test environment that is for most aspects, representative of the production environ-ment ('as-if production').
The goal of acceptance testing is to establish confidence in the system, part of the system or specific non-functional characteristics, e.g. usability, of the system. Acceptance testing is most often focused on a validation type of testing, whereby we are trying to determine whether the system is fit for purpose. Finding defects should not be the main focus in acceptance testing. Although it assesses the system's readiness for deployment and use, it is not necessarily the final level of testing. For example, a large-scale system integration test may come after the acceptance of a system.
Acceptance testing may occur at more than just a single level, for example:
• A Commercial Off The Shelf (COTS) software product may be acceptance tested when it is installed or integrated.
• Acceptance testing of the usability of a component may be done during com ponent testing.
• Acceptance testing of a new functional enhancement may come before system testing.
Within the acceptance test for a business-supporting system, two main test types can be distinguished; as a result of their special character, they are usually prepared and executed separately. The user acceptance test focuses mainly on the functionality thereby validating the fitness-for-use of the system by the business user, while the operational acceptance test (also called production acceptance test) validates whether the system meets the requirements for operation. The user acceptance test is performed by the users and application managers. In terms of planning, the user acceptance test usually links tightly to the system test and will, in many cases, be organ-ized partly overlapping in time. If the system to be tested consists of a number of more or less independent subsystems, the acceptance test for a subsystem that complies to the exit criteria of the system test can start while another subsystem may still be in the system test phase. In most organiza-tions, system administration will perform the operational acceptance test shortly before the system is released. The operational acceptance test may include testing of backup/restore, disaster recovery, maintenance tasks and periodic check of security vulnerabilities.
Other types of acceptance testing that exist are contract acceptance testing and compliance acceptance testing. Contract acceptance testing is performed against a contract's acceptance criteria for producing custom-developed soft-ware. Acceptance should be formally defined when the contract is agreed. Compliance acceptance testing or regulation acceptance testing is performed against the regulations which must be adhered to, such as governmental, legal or safety regulations.
If the system has been developed for the mass market, e.g. commercial off-the-shelf software (COTS), then testing it for individual users or customers is not practical or even possible in some cases. Feedback is needed from potential or existing users in their market before the software product is put out for sale commercially. Very often this type of system undergoes two stages of accept-ance test. The first is called alpha testing. This test takes place at the devel-oper's site. A cross-section of potential users and members of the developer's organization are invited to use the system. Developers observe the users and note problems. Alpha testing may also be carried out by an independent test team. Beta testing, or field testing, sends the system to a cross-section of users who install it and use it under real-world working conditions. The users send records of incidents with the system to the development organization where the defects are repaired.
Note that organizations may use other terms, such as factory acceptance testing and site acceptance testing for systems that are tested before and after being moved to a customer's site.
2.3 TEST TYPES: THE TARGETS OF TESTING
1 Compare four software test types (functional, non-functional, structural and change-related) by example. (K2)
2 Recognize that functional and structural tests occur at any test level. (Kl)
3 Identify and describe non-functional test types based on non-functional requirements. (K2)
4 Identify and describe test types based on the analysis of a software system's structure or architecture. (K2)
5 Describe the purpose of confirmation testing and regression testing.
(K2)
Test types are introduced as a means of clearly defining the objective of a certain test level for a programme or project. We need to think about differ-ent types of testing because testing the functionality of the component or system may not be sufficient at each level to meet the overall test objectives. Focusing the testing on a specific test objective and, therefore, selecting the appropriate type of test helps making and communicating decisions against test objectives easier.
A test type is focused on a particular test objective, which could be the testing of a function to be performed by the component or system; a non-functional quality characteristic, such as reliability or usability; the structure or architecture of the component or system; or related to changes, i.e. con-firming that defects have been fixed (confirmation testing, or re-testing) and looking for unintended changes (regression testing). Depending on its objec-tives, testing will be organized differently. For example, component testing aimed at performance would be quite different to component testing aimed at achieving decision coverage.
2.3.1 Testing of function (functional testing)
The function of a system (or component) is 'what it does'. This is typically described in a requirements specification, a functional specification, or in use cases. There may be some functions that are 'assumed' to be provided that are not documented that are also part of the requirement for a system, though it is difficult to test against undocumented and implicit requirements. Functional tests are based on these functions, described in documents or understood by the testers and may be performed at all test levels (e.g. test for components may be based on a component specification).
Functional testing considers the specified behavior and is often also referred to as black-box testing. This is not entirely true, since black-box testing also includes non-functional testing (see Section 2.3.2).
Function (or functionality) testing can, based upon ISO 9126, be done focus-ing on suitability, interoperability, security, accuracy and compliance. Security testing, for example, investigates the functions (e.g. a firewall) relating to detec-tion of threats, such as viruses, from malicious outsiders.
Testing functionality can be done from two perspectives: requirements-based or business-process-based.
Requirements-based testing uses a specification of the functional require-ments for the system as the basis for designing tests. A good way to start is to use the table of contents of the requirements specification as an initial test inventory or list of items to test (or not to test). We should also prioritize the requirements based on risk criteria (if this is not already done in the specifica-tion) and use this to prioritize the tests. This will ensure that the most impor-tant and most critical tests are included in the testing effort.
Business-process-based testing uses knowledge of the business processes. Business processes describe the scenarios involved in the day-to-day business use of the system. For example, a personnel and payroll system may have a busi-ness process along the lines of: someone joins the company, he or she is paid on a regular basis, and he or she finally leaves the company. Use cases originate from object-oriented development, but are nowadays popular in many develop-ment life cycles. They also take the business processes as a starting point, although they start from tasks to be performed by users. Use cases are a very useful basis for test cases from a business perspective.
The techniques used for functional testing are often specification-based, but experienced-based techniques can also be used (see Chapter 4 for more on test techniques). Test conditions and test cases are derived from the functionality of the component or system. As part of test designing, a model may be developed, such as a process model, state transition model or a plain-language specification.
2.3.2 Testing of software product characteristics (non-functional testing)
A second target for testing is the testing of the quality characteristics, or non-functional attributes of the system (or component or integration group). Here we are interested in how well or how fast something is done. We are testing something that we need to measure on a scale of measurement, for example time to respond.
Non-functional testing, as functional testing, is performed at all test levels. Non-functional testing includes, but is not limited to, performance testing, load testing, stress testing, usability testing, maintainability testing, reliability testing and portability testing. It is the testing of 'how well' the system works.
Many have tried to capture software quality in a collection of characteristics and related sub-characteristics. In these models some elementary characteris-tics keep on reappearing, although their place in the hierarchy can differ. The International Organization for Standardization (ISO) has defined a set of quality characteristics [ISO/IEC 9126, 2001]. This set reflects a major step towards consensus in the IT industry and thereby addresses the general notion of software quality. The ISO 9126 standard defines six quality characteristics and the subdivision of each quality characteristic into a number of
sub-characteristics. This standard is getting more and more recognition in the industry, enabling development, testing and their stakeholders to use a common terminology for quality characteristics and thereby for non-functional testing.
The characteristics and their sub-characteristics are, respectively:
• functionality, which consists of five sub-characteristics: suitability, accuracy, security, interoperability and compliance; this characteristic deals with func tional testing as described in Section 2.3.1;
• reliability, which is defined further into the sub-characteristics maturity (robustness), fault-tolerance, recoverability and compliance;
• usability, which is divided into the sub-characteristics understandability, learnability, operability, attractiveness and compliance;
• efficiency, which is divided into time behavior (performance), resource uti lization and compliance;
• maintainability, which consists of five sub-characteristics: analyzability, changeability, stability, testability and compliance;
• portability, which also consists of five sub-characteristics: adaptability, installability, co-existence, replaceability and compliance.
2.3.3 Testing of software structure/architecture (structural testing)
The third target of testing is the structure of the system or component. If we are talking about the structure of a system, we may call it the system architecture. Structural testing is often referred to as 'white-box' or 'glass-box' because we are interested in what is happening 'inside the box'.
Structural testing is most often used as a way of measuring the thoroughness of testing through the coverage of a set of structural elements or coverage items. It can occur at any test level, although is it true to say that it tends to be mostly applied at component and integration and generally is less likely at higher test levels, except for business-process testing. At component integration level it may be based on the architecture of the system, such as a calling hierar-chy. A system, system integration or acceptance testing test basis could be a business model or menu structure.
At component level, and to a lesser extent at component integration testing, there is good tool support to measure code coverage. Coverage measurement tools assess the percentage of executable elements (e.g. state-ments or decision outcomes) that have been exercised (i.e. covered) by a test suite. If coverage is not 100%, then additional tests may need to be written and run to cover those parts that have not yet been exercised. This of course depends on the exit criteria. (Coverage techniques are covered in Chapter 4.)
The techniques used for structural testing are structure-based techniques, also referred to as white-box techniques. Control flow models are often used to support structural testing.
2.3.4 Testing related to changes (confirmation and regression testing)
The final target of testing is the testing of changes. This category is slightly dif-ferent to the others because if you have made a change to the software, you will have changed the way it functions, the way it performs (or both) and its struc-ture. However we are looking here at the specific types of tests relating to changes, even though they may include all of the other test types.
Confirmation testing (re-testing)
When a test fails and we determine that the cause of the failure is a software defect, the defect is reported, and we can expect a new version of the software that has had the defect fixed. In this case we will need to execute the test again to confirm that the defect has indeed been fixed. This is known as confirmation testing (also known as re-testing).
When doing confirmation testing, it is important to ensure that the test is executed in exactly the same way as it was the first time, using the same inputs, data and environment. If the test now passes does this mean that the software is now correct? Well, we now know that at least one part of the software is correct - where the defect was. But this is not enough. The fix may have intro-duced or uncovered a different defect elsewhere in the software. The way to detect these 'unexpected side-effects' of fixes is to do regression testing.
Regression testing
Like confirmation testing, regression testing involves executing test cases that have been executed before. The difference is that, for regression testing, the test cases probably passed the last time they were executed (compare this with the test cases executed in confirmation testing - they failed the last time).
The term 'regression testing' is something of a misnomer. It would be better if it were called 'anti-regression' testing because we are executing tests with the intent of checking that the system has not regressed (that is, it does not now have more defects in it as a result of some change). More specifically, the purpose of regression testing is to verify that modifications in the software or the environment have not caused unintended adverse side effects and that the system still meets its requirements.
It is common for organizations to have what is usually called a regression test suite or regression test pack. This is a set of test cases that is specifically used for regression testing. They are designed to collectively exercise most functions (cer-tainly the most important ones) in a system but not test any one in detail. It is appropriate to have a regression test suite at every level of testing (component testing, integration testing, system testing, etc.). All of the test cases in a regression test suite would be executed every time a new version of software is produced and this makes them ideal candidates for automation. If the regression test suite is very large it may be more appropriate to select a subset for execution.
Regression tests are executed whenever the software changes, either as a result of fixes or new or changed functionality. It is also a good idea to execute them when some aspect of the environment changes, for example when a new version of a database management system is introduced or a new version of a source code compiler is used.
Maintenance of a regression test suite should be carried out so it evolves over time in line with the software. As new functionality is added to a system new regression tests should be added and as old functionality is changed or removed so too should regression tests be changed or removed. As new tests are added a regression test suite may become very large. If all the tests have to be executed manually it may not be possible to execute them all every time the regression suite is used. In this case a subset of the test cases has to be chosen. This selection should be made in light of the latest changes that have been made to the software. Sometimes a regression test suite of automated tests can become so large that it is not always possible to execute them all. It may be pos-sible and desirable to eliminate some test cases from a large regression test suite for example if they are repetitive (tests which exercise the same condi-tions) or can be combined (if they are always run together). Another approach is to eliminate test cases that have not found a defect for a long time (though this approach should be used with some care!).
2.4 MAINTENANCE TESTING
1 Compare maintenance testing (testing an operational system) to testing a new application with respect to test types, triggers for testing and amount of testing. (K2)
2 Identify reasons for maintenance testing (modifications, migration and retirement). (K2)
3 Describe the role of regression testing and impact analysis in mainte nance. (K2)
Once deployed, a system is often in service for years or even decades. During this time the system and its operational environment is often corrected, changed or extended. Testing that is executed during this life cycle phase is called 'maintenance testing'.
Note that maintenance testing is different from maintainability testing, which defines how easy it is to maintain the system.
The development and test process applicable to new developments does not change fundamentally for maintenance purposes. The same test process steps will apply and, depending on the size and risk of the changes made, several levels of testing are carried out: a component test, an integration test, a system test and an acceptance test. A maintenance test process usually begins with the receipt of an application for a change or a release plan. The test manager will use this as a basis for producing a test plan. On receipt of the new or changed specifications, corresponding test cases are specified or adapted. On receipt of the test object, the new and modified tests and the regression tests are executed. On completion of the testing, the testware is once again preserved.
Comparing maintenance testing to testing a new application is merely a matter of an approach from a different angle, which gives rise to a number of
changes in emphasis. There are several areas where most differences occur, for example regarding the test basis. A 'catching-up' operation is frequently required when systems are maintained. Specifications are often 'missing', and a set of testware relating to the specifications simply does not exist. It may well be possible to carry out this catching-up oper-ation along with testing a new maintenance release, which may reduce the cost. If it is impossible to compile any specifications from which test cases can be written, including expected results, an alternative test basis, e.g. a test oracle, should be sought by way of compromise. A search should be made for documentation which is closest to the specifications and which can be managed by developers as well as testers. In such cases it is advis-able to draw the customer's attention to the lower test quality which may be achieved. Be aware of possible problems of 'daily production'. In the worst case nobody knows what is being tested, many test cases are execut-ing the same scenario and if an incident is found it is often hard to trace it back to the actual defect since no traceability to test designs and/or requirements exists. Note that reproducibility of tests is also important for maintenance testing.
One aspect which, in many cases, differs somewhat from the development situation is the test organization. New development and their appropriate test activities are usually carried out as parts of a project, whereas maintenance tests are normally executed as an activity in the regular organization. As a result, there is often some lack of resources and flexibility, and the test process may experience more competition from other activities.
2.4.1 Impact analysis and regression testing
Usually maintenance testing will consist of two parts:
• testing the changes
• regression tests to show that the rest of the system has not been affected by the maintenance work.
In addition to testing what has been changed, maintenance testing includes extensive regression testing to parts of the system that have not been changed. A major and important activity within maintenance testing is impact analysis. During impact analysis, together with stakeholders, a deci-sion is made on what parts of the system may be unintentionally affected and therefore need careful regression testing. Risk analysis will help to decide where to focus regression testing - it is unlikely that the team will have time to repeat all the existing tests.
If the test specifications from the original development of the system are kept, one may be able to reuse them for regression testing and to adapt them for changes to the system. This may be as simple as changing the expected results for your existing tests. Sometimes additional tests may need to be built. Extension or enhancement to the system may mean new areas have been spec-ified and tests would be drawn up just as for the development. It is also possi-ble that updates are needed to an automated test set, which is often used to support regression testing.
2.4.2 Triggers for maintenance testing
As stated maintenance testing is done on an existing operational system. It is triggered by modifications, migration, or retirement of the system. Modifications include planned enhancement changes (e.g. release-based), cor-rective and emergency changes, and changes of environment, such as planned operating system or database upgrades, or patches to newly exposed or discov-ered vulnerabilities of the operating system. Maintenance testing for migration (e.g. from one platform to another) should include operational testing of the new environment, as well as the changed software. Maintenance testing for the retirement of a system may include the testing of data migration or archiving, if long data-retention periods are required.
Since modifications are most often the main part of maintenance testing foi most organizations, this will be discussed in more detail. From the point of view of testing, there are two types of modifications. There are modifications in which testing may be planned, and there are ad-hoc corrective modifications, which cannot be planned at all. Ad-hoc corrective maintenance takes place when the search for solutions to defects cannot be delayed. Special test proce-dures are required at that time.
Planned modifications
The following types of planned modification may be identified:
• perfective modifications (adapting software to the user's wishes, for instance by supplying new functions or enhancing performance);
• adaptive modifications (adapting software to environmental changes such as new hardware, new systems software or new legislation);
• corrective planned modifications (deferrable correction of defects).
The standard structured test approach is almost fully applicable to planned modifications. On average, planned modification represents over 90% of all maintenance work on systems. [Pol and van Veenendaal]
Ad-hoc corrective modifications
Ad-hoc corrective modifications are concerned with defects requiring an imme-diate solution, e.g. a production run which dumps late at night, a network that goes down with a few hundred users on line, a mailing with incorrect addresses. There are different rules and different procedures for solving problems of this kind. It will be impossible to take the steps required for a structured approach to testing. If, however, a number of activities are carried out prior to a possible malfunction, it may be possible to achieve a situation in which reliable tests car. be executed in spite of 'panic stations' all round. To some extent this type of maintenance testing is often like first aid - patching up - and at a later stage the standard test process is then followed to establish a robust fix, test it and estab-lish the appropriate level of documentation.
A risk analysis of the operational systems should be performed in order to establish which functions or programs constitute the greatest risk to the opera-tional services in the event of disaster. It is then established - in respect of the functions at risk - which (test) actions should be performed if a particular mal-function occurs. Several types of malfunction may be identified and there are
various ways of responding to them for each function at risk. A possible reac-tion might be that a relevant function at risk should always be tested, or that, under certain circumstances, testing might be carried out in retrospect (the next day, for instance). If it is decided that a particular function at risk should always be tested whenever relevant, a number of standard tests, which could be exe-cuted almost immediately, should be prepared for this purpose. The standard tests would obviously be prepared and maintained in accordance with the struc-tured test approach.
Even in the event of ad-hoc modifications, it is therefore possible to bring about an improvement in quality by adopting a specific test approach. It is important to make a thorough risk analysis of the system and to specify a set of standard tests accordingly.
CHAPTER REVIEW
Let's review what you have learned in this chapter.
From Section 2.1, you should now understand the relationship between development and testing within a development life cycle, including the test activities and test (work) products. You should know that the development model to use should fit, or must be adapted to fit, the project and product characteristics. You should be able to recall the reasons for different levels of testing and characteristics of good testing in any life cycle model. You should know the glossary terms (commercial) off-the-shelf software (COTS), incremental development model, test level, validation, verification and V-model.
From Section 2.2, you should know the typical levels of testing. You should be able to compare the different levels of testing with respect to their major objectives, typical objects of testing, typical targets of testing (e.g. functional or structural) and related work products. You should also know which persons perform the testing activities at the various test levels, the types of defects found and failures to be identified. You should know the glossary terms alpha testing, beta testing, component testing, driver, functional requirements, integration, integration testing, non-functional testing, operational testing, regulation acceptance testing (compliance testing), robustness testing, stub, system testing, test-driven development, test environment and user acceptance testing.
From Section 2.3, you should know the four major types of test (functional, non-functional, structural and change-related) and should be able to provide some concrete examples for each of these. You should understand that functional and structural tests occur at any test level and be able to explain how they are applied in the various test levels. You should be able to identify and describe non-functional test types based on non-functional requirements and product quality characteristics. Finally you should be able to explain the purpose of confirmation testing (re-testing) and regression testing in the context of change-related testing. You should know the glossary terms black-box testing, code coverage, confirmation testing (re-testing), functional testing, interoperability testing, load testing, maintainability testing, performance testing, portability testing, regression testing, reliability testing, security testing, specification-based testing, stress testing, structural testing, test suite, usability testing and white-box testing
From Section 2.4, you should be able to compare maintenance testing to testing of new applications. You should be able to identify triggers and reasons for maintenance testing, such as modifications, migration and retirement. Finally you should be able to describe the role of regression testing and impact analysis within maintenance testing. You should know the glossary terms impact analysis and maintenance testing.
SAMPLE EXAM QUESTIONS
Question 1 What are good practices for testing within the development life cycle?
a. Early test analysis and design.
b. Different test levels are defined with specific objectives.
c. Testers will start to get involved as soon as coding is done.
d. A and B above.
Question 2 Which option best describes objec-tives for test levels with a life cycle model?
a. Objectives should be generic for any test level.
b. Objectives are the same for each test level.
c. The objectives of a test level don't need to be defined in advance.
d. Each level has objectives specific to that level.
Question 3 Which of the following is a test type?
a. Component testing
b. Functional testing
c. System testing
d. Acceptance testing
Question 4 Which of the following is a non-functional quality characteristic?
a. Feasibility
b. Usability
c. Maintenance
d. Regression
Question 5 Which of these is a functional test?
a. Measuring response time on an on-line booking system.
b. Checking the effect of high volumes of traffic in a call-center system.
c. Checking the on-line bookings screen informa tion and the database contents against the infor mation on the letter to the customers.
d. Checking how easy the system is to use.
Question 6 Which of the following is a true statement regarding the process of fixing emergency changes?
a. There is no time to test the change before it goes live, so only the best developers should do this work and should not involve testers as they slow down the process.
b. Just run the retest of the defect actually fixed.
c. Always run a full regression test of the whole system in case other parts of the system have been adversely affected.
d. Retest the changed area and then use risk assessment to decide on a reasonable subset of the whole regression test to run in case other parts of the system have been adversely affected.
Question 7 A regression test:
a. Is only run once.
b. Will always be automated.
c. Will check unchanged areas of the software to see if they have been affected.
d. Will check changed areas of the software to see if they have been affected.
Question 8 Non-functional testing includes:
a. Testing to see where the system does not func tion correctly.
b. Testing the quality attributes of the system including reliability and usability.
c. Gaining user approval for the system.
d. Testing a system feature using only the software required for that function.
Question 9 Beta testing is:
a. Performed by customers at their own site.
b. Performed by customers at the software devel oper's site.
c. Performed by an independent test team.
d. Useful to test software developed for a specific customer or user.
CHAPTER 3 Static techniques
tatic test techniques provide a powerful way to improve the quality and productivity of software development. This chapter describes static test techniques, including reviews, and provides an overview of how they are conducted. The fundamental objective of static testing is to improve the quality of software work products by assisting engineers to recognize and fix their own defects early in the software development process. While static testing techniques will not solve all the problems, they are enormously effective. Static techniques can improve both quality and productivity by impressive factors. Static testing is not magic and it should not be considered a replacement for dynamic testing, but all software organizations should consider using reviews in all major aspects of their work including requirements, design, implementation, testing, and maintenance. Static analysis tools implement automated checks, e.g. on code.
3.1 REVIEWS AND THE TEST PROCESS
1 Recognize software work products that can be examined by different static techniques. (Kl)
2 Describe the importance and value of considering static techniques for the assessment of software work products. (K2)
3 Explain the difference between static and dynamic techniques. (K2)
In Chapter 1, several testing terms were presented. Also testing itself was defined. The latter definition is repeated here as a means for explaining the two major types of testing.
The definition of testing outlines objectives that relate to evaluation, revealing defects and quality. As indicated in the definition two approaches can be used to achieve these objectives, static testing and dynamic testing.
With dynamic testing methods, software is executed using a set of input values and its output is then examined and compared to what is expected. During static testing, software work products are examined manually, or with a set of tools, but not executed. As a consequence, dynamic testing can only be applied to software code. Dynamic execution is applied as a technique to detect
defects and to determine quality attributes of the code. This testing option is not applicable for the majority of the software work products. Among the ques-tions that arise are: How can we evaluate or analyze a requirements document, a design document, a test plan, or a user manual? How can we effectively pre-examine the source code before execution? One powerful technique that can be used is static testing, e.g. reviews. In principle all software work products can be tested using review techniques.
Dynamic testing and static testing are complementary methods, as they tend to find different types of defects effectively and efficiently. Types of defects that are easier to find during static testing are: deviations from standards, missing requirements, design defects, non-maintainable code and inconsistent interface specifications. Note that in contrast to dynamic testing, static testing finds defects rather than failures.
In addition to finding defects, the objectives of reviews are often also informational, communicational and educational, whereby participants learn about the content of software work products to help them understand the role of their own work and to plan for future stages of development. Reviews often represent project milestones, and support the establishment of a baseline for a software product. The type and quantity of defects found during reviews can also help testers focus their testing and select effective classes of tests. In some cases customers/users attend the review meeting and provide feedback to the development team, so reviews are also a means of customer/user communication.
Studies have shown that as a result of reviews, a significant increase in pro-ductivity and product quality can be achieved [Gilb and Graham, 1993], [van Veenendaal, 1999]. Reducing the number of defects early in the product life cycle also means that less time has to be spent on testing and maintenance. To summarize, the use of static testing, e.g. reviews, on software work products has various advantages:
• Since static testing can start early in the life cycle, early feedback on quality issues can be established, e.g. an early validation of user requirements and not just late in the life cycle during acceptance testing.
• By detecting defects at an early stage, rework costs are most often relatively low and thus a relatively cheap improvement of the quality of software prod ucts can be achieved.
• Since rework effort is substantially reduced, development productivity figures are likely to increase.
• The evaluation by a team has the additional advantage that there is an exchange of information between the participants.
• Static tests contribute to an increased awareness of quality issues.
In conclusion, static testing is a very suitable method for improving the quality of software work products. This applies primarily to the assessed products themselves, but it is also important that the quality improvement is not achieved once but has a more structural character. The feedback from the static testing process to the development process allows for process improvement, which supports the avoidance of similar errors being made in the future.
3,2 REVIEW PROCESS
1 Recall the phases, roles and responsibilities of a typical formal review.
(Kl)
2 Explain the differences between different types of review: informal review, technical review, walkthrough and inspection. (K2)
3 Explain the factors for successful performance of reviews. (K2)
Reviews vary from very informal to formal (i.e. well structured and regulated). Although inspection is perhaps the most documented and formal review tech-nique, it is certainly not the only one. The formality of a review process is related to factors such as the maturity of the development process, any legal or regulatory requirements or the need for an audit trail. In practice the informal review is perhaps the most common type of review. Informal reviews are applied at various times during the early stages in the life cycle of a document. A two-person team can conduct an informal review, as the author can ask a col-league to review a document or code. In later stages these reviews often involve more people and a meeting. This normally involves peers of the author, who try to find defects in the document under review and discuss these defects in a review meeting. The goal is to help the author and to improve the quality of the document. Informal reviews come in various shapes and forms, but all have one characteristic in common - they are not documented.
3.2.1 Phases of a formal review
In contrast to informal reviews, formal reviews follow a formal process. A typical formal review process consists of six main steps:
1 Planning
2 Kick-off
3 Preparation
4 Review meeting
5 Rework
6 Follow-up.
Planning
The review process for a particular review begins with a 'request for review' by the author to the moderator (or inspection leader). A moderator is often assigned to take care of the scheduling (dates, time, place and invitation) of the review. On a project level, the project planning needs to allow time for review and rework activities, thus providing engineers with time to thoroughly participate in reviews. For more formal reviews, e.g. inspections, the moderator always performs an entry check and defines at this stage formal exit criteria. The entry check is
carried out to ensure that the reviewers' time is not wasted on a document that is not ready for review. A document containing too many obvious mistakes is clearly not ready to enter a formal review process and it could even be very harmful to the review process. It would possibly de-motivate both reviewers and the author. Also, the review is most likely not effective because the numerous obvious and minor defects will conceal the major defects.
Although more and other entry criteria can be applied, the following can be regarded as the minimum set for performing the entry check:
• A short check of a product sample by the moderator (or expert) does not reveal a large number of major defects. For example, after 30 minutes of checking, no more than 3 major defects are found on a single page or fewer than 10 major defects in total in a set of 5 pages.
• The document to be reviewed is available with line numbers.
• The document has been cleaned up by running any automated checks that apply.
• References needed for the inspection are stable and available.
• The document author is prepared to join the review team and feels confident with the quality of the document.
If the document passes the entry check, the moderator and author decide which part of the document to review. Because the human mind can com-prehend a limited set of pages at one time, the number should not be too high. The maximum number of pages depends, among other things, on the objective, review type and document type and should be derived from prac-tical experiences within the organization. For a review, the maximum size is usually between 10 and 20 pages. In formal inspection, only a page or two may be looked at in depth in order to find the most serious defects that are not obvious.
After the document size has been set and the pages to be checked have been selected, the moderator determines, in co-operation with the author, the com-position of the review team. The team normally consists of four to six partici-pants, including moderator and author. To improve the effectiveness of the review, different roles are assigned to each of the participants. These roles help the reviewers focus on particular types of defects during checking. This reduces the chance of different reviewers finding the same defects. The moderator assigns the roles to the reviewers.
Figure 3.1 shows the different roles within a review. The roles represent views of the document under review.
Within reviews the following focuses can be identified:
• focus on higher-level documents, e.g. does the design comply to the requirements;
• focus on standards, e.g. internal consistency, clarity, naming conventions, templates;
• focus on related documents at the same level, e.g. interfaces between soft ware functions;
• focus on usage, e.g. for testability or maintainability.
The author may raise additional specific roles and questions that have to be addressed. The moderator has the option to also fulfil a role, alongside the task of being a review leader. Checking the document improves the moderator's ability to lead the meeting, because it ensures better understanding. Furthermore, it improves the review efficiency because the moderator replaces an engineer that would other-wise have to check the document and attend the meeting. It is recommended that the moderator take the role of checking compliance to standards, since this tends to be a highly objective role, which leads to less discussion of the defects found.
Kick-off
An optional step in a review procedure is a kick-off meeting. The goal of this meeting is to get everybody on the same wavelength regarding the document under review and to commit to the time that will be spent on checking. Also the result of the entry check and defined exit criteria are discussed in case of a more formal review. In general a kick-off is highly recommended since there is a strong positive effect of a kick-off meeting on the motivation of reviewers and thus the effectiveness of the review process. At customer sites, we have meas-ured results up to 70% more major defects found per page as a result of per-forming a kick-off, [van Veenendaal and van der Zwan, 2000]
During the kick-off meeting the reviewers receive a short introduction on the objectives of the review and the documents. The relationships between the doc-ument under review and the other documents (sources) are explained, espe-cially if the number of related documents is high.
Role assignments, checking rate, the pages to be checked, process changes and possible other questions are also discussed during this meeting. Of course the distribution of the document under review, source documents and other related documentation, can also be done during the kick-off.
Preparation
The participants work individually on the document under review using the related documents, procedures, rules and checklists provided. The individual participants identify defects, questions and comments, according to their
understanding of the document and role. All issues are recorded, preferably using a logging form. Spelling mistakes are recorded on the document under review but not mentioned during the meeting. The annotated document will be given to the author at the end of the logging meeting. Using checklists during this phase can make reviews more effective and efficient, for example a specific checklist based on perspectives such as user, maintainer, tester or operations, or a checklist for typical coding problems.
A critical success factor for a thorough preparation is the number of pages checked per hour. This is called the checking rate. The optimum checking rate is the result of a mix of factors, including the type of document, its com-plexity, the number of related documents and the experience of the reviewer. Usually the checking rate is in the range of five to ten pages per hour, but may be much less for formal inspection, e.g. one page per hour. During preparation, participants should not exceed this criterion. By collecting data and measuring the review process, company-specific criteria for checking rate and document size (see planning phase) can be set, preferably specific to a document type.
Review meeting
The meeting typically consists of the following elements (partly depending on
the review type): logging phase, discussion phase and decision phase.
During the logging phase the issues, e.g. defects, that have been identified during the preparation are mentioned page by page, reviewer by reviewer and are logged either by the author or by a scribe. A separate person to do the logging (a scribe) is especially useful for formal review types such as an inspec-tion. To ensure progress and efficiency, no real discussion is allowed during the logging phase. If an issue needs discussion, the item is logged and then handled in the discussion phase. A detailed discussion on whether or not an issue is a defect is not very meaningful, as it is much more efficient to simply log it and proceed to the next one. Furthermore, in spite of the opinion of the team, a dis-cussed and discarded defect may well turn out to be a real one during rework.
Every defect and its severity should be logged. The participant who identifies the defect proposes the severity. Severity classes could be:
• Critical: defects will cause downstream damage; the scope and impact of the defect is beyond the document under inspection.
• Major, defects could cause a downstream effect (e.g. a fault in a design can result in an error in the implementation).
• Minor, defects are not likely to cause downstream damage (e.g. non-compli ance with the standards and templates). ,
In order to keep the added value of reviews, spelling errors are not part of the defect classification. Spelling defects are noted, by the participants, in the document under review and given to the author at the end of the meeting or could be dealt with in a separate proofreading exercise.
During the logging phase the focus is on logging as many defects as possible within a certain timeframe. To ensure this, the moderator tries to keep a good logging rate (number of defects logged per minute). In a well-led and disci-plined formal review meeting, the logging rate should be between one and two defects logged per minute.
For a more formal review, the issues classified as discussion items will be handled during this meeting phase. Informal reviews will often not have a sep-arate logging phase and will start immediately with discussion. Participants can take part in the discussion by bringing forward their comments and rea-soning. As chairman of the discussion meeting, the moderator takes care of people issues. For example, the moderator prevents discussions from getting too personal, rephrases remarks if necessary and calls for a break to cool down 'heated' discussions and/or participants.
Reviewers who do not need to be in the discussion may leave, or stay as a learning exercise. The moderator also paces this part of the meeting and ensures that all discussed items either have an outcome by the end of the meeting, or are defined as an action point if a discussion cannot be solved during the meeting. The outcome of discussions is documented for future reference.
At the end of the meeting, a decision on the document under review has to be made by the participants, sometimes based on formal exit criteria. The most important exit criterion is the average number of critical and/or major defects found per page (e.g. no more than three critical/major defects per page). If the number of defects found per page exceeds a certain level, the document must be reviewed again, after it has been reworked. If the document complies with the exit criteria, the document will be checked during follow-up by the moder-ator or one or more participants. Subsequently, the document can leave the review process.
If a project is under pressure, the moderator will sometimes be forced to skip re-reviews and exit with a defect-prone document. Setting, and agreeing, quantified exit level criteria helps the moderator to make firm decisions at all times.
In addition to the number of defects per page, other exit criteria are used that measure the thoroughness of the review process, such as ensuring that all pages have been checked at the right rate. The average number of defects per page is only a valid quality indicator if these process criteria are met.
Rework
Based on the defects detected, the author will improve the document under review step by step. Not every defect that is found leads to rework. It is the author's responsibility to judge if a defect has to be fixed. If nothing is done about an issue for a certain reason, it should be reported to at least indicate that the author has considered the issue.
Changes that are made to the document should be easy to identify during follow-up. Therefore the author has to indicate where changes are made (e.g. using 'Track changes' in word-processing software).
Follow-up
The moderator is responsible for ensuring that satisfactory actions have been taken on all (logged) defects, process improvement suggestions and change requests. Although the moderator checks to make sure that the author has taken action on all known defects, it is not necessary for the moderator to check all the corrections in detail. If it is decided that all participants will check the updated document, the moderator takes care of the distribution and collects
f
the feedback. For more formal review types the moderator checks for compli-ance to the exit criteria.
In order to control and optimize the review process, a number of measure-ments are collected by the moderator at each step of the process. Examples of such measurements include number of defects found, number of defects found per page, time spent checking per page, total review effort, etc. It is the respon-sibility of the moderator to ensure that the information is correct and stored for future analysis.
3.2.2 Roles and responsibilities
The participants in any type of formal review should have adequate knowledge of the review process. The best, and most efficient, review situation occurs when the participants gain some kind of advantage for their own work during review-ing. In the case of an inspection or technical review, participants should have been properly trained as both types of review have proven to be far less success-ful without trained participants. This indeed is perceived as being a critical success factor.
The best formal reviews come from well-organized teams, guided by trained moderators (or review leaders). Within a review team, four types of participants can be distinguished: moderator, author, scribe and reviewer. In addition man-agement needs to play a role in the review process.
The moderator
The moderator (or review leader) leads the review process. He or she deter-mines, in co-operation with the author, the type of review, approach and the composition of the review team. The moderator performs the entry check and the follow-up on the rework, in order to control the quality of the input and output of the review process. The moderator also schedules the meeting, disseminates documents before the meeting, coaches other team members, paces the meeting, leads possible discussions and stores the data that is collected.
The author
As the writer of the document under review, the author's basic goal should
be to learn as much as possible with regard to improving the quality of the
document, but also to improve his or her ability to write future documents.
The author's task is to illuminate unclear areas and to understand the defects
found.
The scribe
During the logging meeting, the scribe (or recorder) has to record each defect mentioned and any suggestions for process improvement. In practice it is often the author who plays this role, ensuring that the log is readable and understand-able. If authors record their own defects, or at least make their own notes in their own words, it helps them to understand the log better during rework. However, having someone other than the author take the role of the scribe (e.g. the moderator) can have significant advantages, since the author is freed up to think about the document rather than being tied down with lots of writing.
The reviewers
The task of the reviewers (also called checkers or inspectors) is to check any material for defects, mostly prior to the meeting. The level of thoroughness required depends on the type of review. The level of domain knowledge or tech-nical expertise needed by the reviewers also depends on the type of review. Reviewers should be chosen to represent different perspectives and roles in the review process. In addition to the document under review, the material review-ers receive includes source documents, standards, checklists, etc. In general, the fewer source and reference documents provided, the more domain expertise regarding the content of the document under review is needed.
The manager
The manager is involved in the reviews as he or she decides on the execution of reviews, allocates time in project schedules and determines whether review process objectives have been met. The manager will also take care of any review training requested by the participants. Of course a manager can also be involved in the review itself depending on his or her background, playing the role of a reviewer if this would be helpful.
3.2.3 Types of review
A single document may be the subject of more than one review. If more than one type of review is used, the order may vary. For example, an informal review may be carried out before a technical review, or an inspection may be carried out on a requirements specification before a walkthrough with customers. It is apparent that none of the following types of review is the 'winner', but the dif-ferent types serve different purposes at different stages in the life cycle of a document.
The main review types, their main characteristics and common objectives are described below.
Walkthrough
A walkthrough is characterized by the author of the document under review guiding the participants through the document and his or her thought processes, to achieve a common understanding and to gather feedback. This is especially useful if people from outside the software discipline are present, who are not used to, or cannot easily understand software development documents. The content of the document is explained step by step by the author, to reach consensus on changes or to gather information.
Within a walkthrough the author does most of the preparation. The partici-pants, who are selected from different departments and backgrounds, are not required to do a detailed study of the documents in advance. Because of the way the meeting is structured, a large number of people can participate and this larger audience can bring a great number of diverse viewpoints regarding the contents of the document being reviewed as well as serving an educational purpose. If the audience represents a broad cross-section of skills and disci-plines, it can give assurance that no major defects are 'missed' in the walk-through. A walkthrough is especially useful for higher-level documents, such as requirement specifications and architectural documents.
The specific goals of a walkthrough depend on its role in the creation of the document. In general the following goals can be applicable:
• to present the document to stakeholders both within and outside the soft ware discipline, in order to gather information regarding the topic under documentation;
• to explain (knowledge transfer) and evaluate the contents of the docu ment;
• to establish a common understanding of the document;
• to examine and discuss the validity of proposed solutions and the viability of alternatives, establishing consensus.
Key characteristics of walkthroughs are:
• The meeting is led by the authors; often a separate scribe is present.
• Scenarios and dry runs may be used to validate the content.
• Separate pre-meeting preparation for reviewers is optional.
Technical review
A technical review is a discussion meeting that focuses on achieving con-sensus about the technical content of a document. Compared to inspec-tions, technical reviews are less formal and there is little or no focus on defect identification on the basis of referenced documents, intended read-ership and rules. During technical reviews defects are found by experts, who focus on the content of the document. The experts that are needed for a technical review are, for example, architects, chief designers and key users. In practice, technical reviews vary from quite informal to very formal.
The goals of a technical review are to:
• assess the value of technical concepts and alternatives in the product and project environment;
• establish consistency in the use and representation of technical concepts;
• ensure, at an early stage, that technical concepts are used correctly;
• inform participants of the technical content of the document.
Key characteristics of a technical review are:
• It is a documented defect-detection process that involves peers and technical experts.
• It is often performed as a peer review without management partici pation.
• Ideally it is led by a trained moderator, but possibly also by a technical expert.
• A separate preparation is carried out during which the product is examined and the defects are found.
• More formal characteristics such as the use of checklists and a logging list or issue log are optional.
Inspection
Inspection is the most formal review type. The document under inspection is prepared and checked thoroughly by the reviewers before the meeting, compar-ing the work product with its sources and other referenced documents, and using rules and checklists. In the inspection meeting the defects found are logged and any discussion is postponed until the discussion phase. This makes the inspection meeting a very efficient meeting.
The reason for carrying out inspections can be explained by using Weinberg's concept of egoless engineering [Weinberg, 1971]. Weinberg refers to the human tendency to self-justify actions. Since we tend not to see evidence that conflicts with our strong beliefs, our ability to find errors in our own work is impaired. Because of this tendency, many engineering organizations have established independent test groups that specialize in finding defects. Similar principles have led to the introduction of inspections and reviews in general.
Depending on the organization and the objectives of a project, inspections can be balanced to serve a number of goals. For example, if the time to market is extremely important, the emphasis in inspections will be on efficiency. In a safety-critical market, the focus will be on effectiveness.
The generally accepted goals of inspection are to:
• help the author to improve the quality of the document under inspection;
• remove defects efficiently, as early as possible;
• improve product quality, by producing documents with a higher level of quality;
• create a common understanding by exchanging information among the inspection participants;
• train new employees in the organization's development process;
• learn from defects found and improve processes in order to prevent recur rence of similar defects;
• sample a few pages or sections from a larger document in order to measure the typical quality of the document, leading to improved work by individuals in the future, and to process improvements.
Key characteristics of an inspection are:
• It is usually led by a trained moderator (certainly not by the author).
• It uses defined roles during the process.
• It involves peers to examine the product.
• Rules and checklists are used during the preparation phase.
• A separate preparation is carried out during which the product is examined and the defects are found.
• The defects found are documented in a logging list or issue log.
• A formal follow-up is carried out by the moderator applying exit criteria.
• Optionally, a causal analysis step is introduced to address process improve ment issues and learn from the defects found.
• Metrics are gathered and analyzed to optimize the process.
3.2.4 Success factors for reviews
Implementing (formal) reviews is not easy as there is no one way to success and there are numerous ways to fail. The next list contains a number of critical success factors that improve the chances of success when implementing reviews. It aims to answer the question, 'How do you start (formal) reviews?'.
Find a 'champion'
A champion is needed, one who will lead the process on a project or organiza-tional level. They need expertise, enthusiasm and a practical mindset in order to guide moderators and participants. The authority of this champion should be clear to the entire organization. Management support is also essential for success. They should, amongst other things, incorporate adequate time for review activities in project schedules.
Pick things that really count
Select the documents for review that are most important in a project. Reviewing highly critical, upstream documents like requirements and architec-ture will most certainly show the benefits of the review process to the project. These invested review hours will have a clear and high return on investment. In addition make sure each review has a clear objective and the correct type of review is selected that matches the defined objective. Don't try and review everything by inspection; fit the review to the risk associated with the docu-ment. Some documents may only warrant an informal review and others will repay using inspection. Of course it is also of utmost importance that the right people are involved.
Explicitly plan and track review activities
To ensure that reviews become part of the day-to-day activities, the hours to be spent should be made visible within each project plan. The engineers involved are prompted to schedule time for preparation and, very impor-tantly, rework. Tracking these hours will improve planning of the next review. As stated earlier, management plays an important part in planning of review activities.
Train participants
It is important that training is provided in review techniques, especially the more formal techniques, such as inspection. Otherwise the process is likely to be impeded by those who don't understand the process and the reasoning behind it. Special training should be provided to the moderators to prepare them for their critical role in the review process.
Manage people issues
Reviews are about evaluating someone's document. Some reviews tend to get
too personal when they are not well managed by the moderator. People issues
and psychological aspects should be dealt with by the moderator and should
be part of the review training, thus making the review a positive experience
for the author. During the review, defects should be welcomed and expressed
objectively.
Follow the rules but keep it simple
Follow all the formal rules until you know why and how to modify them, but make the process only as formal as the project culture or maturity level allows. Do not become too theoretical or too detailed. Checklists and roles are recom-mended to increase the effectiveness of defect identification.
Continuously improve process and tools
Continuous improvement of process and supporting tools (e.g. checklists),
based upon the ideas of participants, ensures the motivation of the engineers
involved. Motivation is the key to a successful change process. There should
also be an emphasis, in addition to defect finding, on learning and process
improvement.
Report results
Report quantified results and benefits to all those involved as soon as possible, and discuss the consequences of defects if they had not been found this early. Costs should of course be tracked, but benefits, especially when problems don't occur in the future, should be made visible by quantifying the benefits as well as the costs.
Just do it!
The process is simple but not easy. Each step of the process is clear, but expe-rience is needed to execute them correctly. So, try to get experienced people to observe and help where possible. But most importantly, start doing reviews and start learning from every review.
3.3 STATIC ANALYSIS BY TOOLS
1 Describe the objective of static analysis and compare it to dynamic testing. (K2)
2 Recall typical defects identified by static analysis and compare them to reviews and dynamic testing. (Kl)
3 List typical benefits of static analysts. (Kl)
4 List typical code and design defects that may be identified by static analysis tools. (Kl)
There is much to be done examining software work products without actually running the system. For example, we saw in the previous section that we can carefully review requirements, designs, code, test plans and more, to find defects and fix them before we deliver a product to a customer. In this section, we focus on a different kind of static testing, where we carefully examine requirements, designs and code, usually with automated assistance to ferret out
additional defects before the code is actually run. Thus, what is called static analysis is just another form of testing.
Static analysis is an examination of requirements, design and code that differs from more traditional dynamic testing in a number of important ways:
• Static analysis is performed on requirements, design or code without actually executing the software artifact being examined.
• Static analysis is ideally performed before the types of formal review dis cussed in Section 3.2.
• Static analysis is unrelated to dynamic properties of the requirements, design and code, such as test coverage.
• The goal of static analysis is to find defects, whether or not they may cause failures. As with reviews, static analysis finds defects rather than failures.
For static analysis there are many tools, and most of them focus on soft-ware code. Static analysis tools are typically used by developers before, and sometimes during, component and integration testing and by designers during software modeling. The tools can show not only structural attributes (code metrics), such as depth of nesting or cyclomatic number and check against coding standards, but also graphic depictions of control flow, data relationships and the number of distinct paths from one line of code to another. Even the compiler can be considered a static analysis tool, since it builds a symbol table, points out incorrect usage and checks for non-compli-ance to coding language conventions (syntax).
One of the reasons for using static analysis (coding standards and the like) is related to the characteristics of the programming languages themselves. One may think that the languages are safe to use, because at least the stan-dards committee knows where the problems are. But this would be wrong. Adding to the holes left by the standardization process, programmers con-tinue to report features of the language, which though well-defined, lead to recognizable fault modes. By the end of the 1990s, approximately 700 of these additional problems had been identified in standard C. It is now clear that such fault modes exist. It can be demonstrated that they frequently escape the scrutiny of conventional dynamic testing, ending up in commer-cial products. These problems can be found by using static analysis tools to detect them. In fact, many of the 700 fault modes reported in C can be detected in this way! In a typical C program, there is an average of approxi-mately eight such faults per 1000 lines of source code; they are embedded in the released code, just waiting to cause the code to fail [Hatton, 1997]. Dynamic testing simply did not detect them. C is not the culprit here; this exercise can be carried out for other languages with broadly the same results. All programming languages have problems and programmers cannot assume that they are protected against them. And nothing in the current interna-tional process of standardizing languages will prevent this from happening in the future.
The various features of static analysis tools are discussed below, with a special focus toward static code analysis tools since these are the most common in day-to-day practice. Note that static analysis tools analyze software code, as well as generated output such as HTML and XML.
3.3.1 Coding standards
Checking for adherence to coding standards is certainly the most well-known of all features. The first action to be taken is to define or adopt a coding stan-dard. Usually a coding standard consists of a set of programming rules (e.g. 'Always check boundaries on an array when copying to that array'), naming conventions (e.g. 'Classes should start with capital C) and layout specifica-tions (e.g. 'Indent 4 spaces'). It is recommended that existing standards are adopted. The main advantage of this is that it saves a lot of effort. An extra reason for adopting this approach is that if you take a well-known coding stan-dard there will probably be checking tools available that support this standard. It can even be put the other way around: purchase a static code analyzer and declare (a selection of) the rules in it as your coding standard. Without such tools, the enforcement of a coding standard in an organization is likely to fail. There are three main causes for this: the number of rules in a coding standard is usually so large that nobody can remember them all; some context-sensitive rules that demand reviews of several files are very hard to check by human beings; and if people spend time checking coding standards in reviews, that will distract them from other defects they might otherwise find, making the review process less effective.
3.3.2 Code metrics
As stated, when performing static code analysis, usually information is calculated about structural attributes of the code, such as comment fre-quency, depth of nesting, cyclomatic number and number of lines of code. This information can be computed not only as the design and code are being created but also as changes are made to a system, to see if the design or code is becoming bigger, more complex and more difficult to understand and maintain. The measurements also help us to decide among several design alternatives, especially when redesigning portions of existing code.
There are many different kinds of structural measures, each of which tells us something about the effort required to write the code in the first place, to understand the code when making a change, or to test the code using particu-lar tools or techniques.
Experienced programmers know that 20% of the code will cause 80% of the problems, and complexity analysis helps to find that all-important 20%, which relate back to the principle on defect clustering as explained in Chapter 1. Complexity metrics identify high risk, complex areas.
The cyclomatic complexity metric is based on the number of decisions in a program. It is important to testers because it provides an indication of the amount of testing (including reviews) necessary to practically avoid defects. In other words, areas of code identified as more complex are can-didates for reviews and additional dynamic tests. While there are many ways to calculate cyclomatic complexity, the easiest way is to sum the number of binary decision statements (e.g. if, while, for, etc.) and add 1 to it. A more formal definition regarding the calculation rules is provided in the glossary.
Below is a simple program as an example:
IF A = 354
THEN IF B > C
THEN A = B
ELSE A = C
ENDIF
ENDIF
Print A
The control flow generated from the program would look like Figure 3.2.
The control flow shows seven nodes (shapes) and eight edges (lines), thus using the formal formula the cyclomatic complexity is 8-7 + 2 = 3. In this case there is no graph called or subroutine. Alternatively one may calculate the cyclomatic complexity using the decision points rule. Since there are two decision points, the cyclomatic complexity is 2 + 1 = 3.
3.3.3 Code structure
There are many different kinds of structural measures, each of which tells us something about the effort required to write the code in the first place, to understand the code when making a change, or to test the code using particular tools or techniques. It is often assumed that a large module takes longer to specify, design, code and test than a smaller one. But in fact the code's structure plays a big part. There are several aspects of code structure to consider:
• control flow structure;
• data flow structure;
• data structure.
The control flow structure addresses the sequence in which the instructions are executed. This aspect of structure reflects the iterations and loops in a program's design. If only the size of a program is measured, no information is provided on how often an instruction is executed as it is run. Control flow analysis can also be used to identify unreachable (dead) code. In fact many of the code metrics relate to the control flow structure, e.g. number of nested levels or cyclomatic complexity.
Data flow structure follows the trail of a data item as it is accessed and mod-ified by the code. Many times, the transactions applied to data are more complex than the instructions that implement them. Thus, using data flow measures it is shown how the data act as they are transformed by the program. Defects can be found such as referencing a variable with an undefined value and variables that are never used.
Data structure refers to the organization of the data itself, independent of the program. When data is arranged as a list, queue, stack, or other well-defined structure, the algorithms for creating, modifying or deleting them are more likely to be well-defined, too. Thus, the data structure provides a lot of informa-tion about the difficulty in writing programs to handle the data and in designing test cases to show program correctness. That is, sometimes a program is complex because it has a complex data structure, rather than because of complex control or data flow.
The important thing for the tester is to be aware that the above mentioned static analysis measures can be used as early warning signals of how good the code is likely to be when it is finished.
In summary the value of static analysis is especially for:
• early detection of defects prior to test execution;
• early warning about suspicious aspects of the code, design or requirements;
• identification of defects not easily found in dynamic testing;
• improved maintainability of code and design since engineers work according to documented standards and rules;
• prevention of defects, provided that engineers are willing to learn from their errors and continuous improvement is practised.
CHAPTER REVIEW
Let's review what you have learned in this chapter.
From Section 3.1, you should be able to explain the importance and advan-tages of static testing. You should know the difference between static testing and dynamic testing, and also understand the concept of reviews. You should be able to recognize the software work products that can be examined by static testing. You should know the glossary terms static testing, dynamic testing and reviews.
From Section 3.2, you should understand the difference between formal and informal reviews. You should be able to recall the main phases of a typical formal review. The main roles within reviews and their responsibilities should be clear to you. You should know the differences between the various types of formal review: technical review, walkthrough and inspection. Finally you should be able to explain the factors for successful performance of reviews. You should know the glossary terms entry criteria, exit criteria, formal review, informal review, inspection, moderator, reviewer, scribe, technical review and walkthrough.
From Section 3.3, you should understand the objective of static analysis and be able to compare it to static and dynamic testing. You should be able to describe the main features of static analysis and recall typical defects that can be found using static analysis. Finally, you should be able to recall the benefits of using static analysis. You should know the glossary terms compiler, cyclo-matic complexity, control flow, data flow and static analysis.
SAMPLE EXAM QUESTIONS
Question 1 Which of the following artifacts can be examined by using review techniques?
a. Software code
b. Requirements specification
c. Test designs
d. All of the above
Question 2 Which statement about the function of a static analysis tool is true?
a. Gives quality information about the code without executing it.
b. Checks expected results against actual results.
c. Can detect memory leaks.
d. Gives information about what code has and has not been exercised.
Question 3 Which is not a type of review?
a. Walkthrough
b. Inspection
c. Informal review
d. Management approval
Question 4 What statement about reviews is true?
a. Inspections are led by a trained moderator, whereas technical reviews are not necessarily.
b. Technical reviews are led by a trained leader, inspections are not.
c. In a walkthrough, the author does not attend.
d. Participants for a walkthrough always need to be thoroughly trained.
Question 5 What is the main difference between a walkthrough and an inspection?
a. An inspection is led by the authors, whilst a walk through is led by a trained moderator.
b. An inspection has a trained leader, whilst a walk through has no leader.
c. Authors are not present during inspections, whilst they are during walkthroughs.
d. A walkthrough is led by the author, whilst an inspection is led by a trained moderator.
Question 6 Which of the following characteristics and types of review processes belong together?
1. Led by the author
2. Undocumented
3. No management participation
4. Led by a trained moderator or leader
5. Uses entry and exit criteria
s. Inspection
t. Technical review
u. Informal review
v. Walkthrough
a. s = 4, t = 3, u = 2 and 5, v = 1
b. s = 4 and 5, t = 3, u = 2, v = 1
c. s = 1 and 5, t = 3, u = 2, v = 4
d. s = 5, t = 4, u = 3, v = 1 and 2
Question 7 What statement about static analysis is true?
a. With static analysis, defects can be found that are difficult to find with dynamic testing.
b. Compiling is not a form of static analysis.
c. When properly performed, static analysis makes functional testing redundant.
d. Static analysis finds all faults.
Question 8 Which of the following statements about early test design are true and which are false?
1. Defects found during early test design are more expensive to fix.
2. Early test design can find defects.
3. Early test design can cause changes to the requirements.
4. Early test design takes more effort.
a. 1 and 3 are true. 2 and 4 are false.
b. 2 is true. 1, 3 and 4 are false.
c. 2 and 3 are true. 1 and 4 are false.
d. 2, 3 and 4 are true. 1 is false.
Question 9 Static code analysis typically identifies all but one of the following problems. Which is it?
a. Unreachable code
b. Undeclared variables
c. Faults in the requirements
d. Too few comments
CHAPTER 4 - Test design techniques
hapter 3 covered static testing, looking at documents and code, but not running the code we are interested in. This chapter looks at dynamic testing, where the software we are interested in is run by executing tests on the running code.
4.1 IDENTIFYING TEST CONDITIONS AND DESIGNING TEST CASES
1 Differentiate between a test design specification, a test case specification and a test procedure specification. (Kl)
2 Compare the terms test condition, test case and test procedure. (K2)
3 Write test cases: (K3)
a showing a clear traceability to the requirements; b containing an expected result.
4 Translate test cases into a well-structured test procedure specification at a level of detail relevant to the knowledge of the testers. (K3)
5 Write a test execution schedule for a given set of test cases, considering prioritization, and technical and logical dependencies. (K3)
4.1.1 Introduction
Before we can actually execute a test, we need to know what we are trying to test, the inputs, the results that should be produced by those inputs, and how we actually get ready for and run the tests.
In this section we are looking at three things: test conditions, test cases and test procedures (or scripts) - they are described in the sections below. Each is specified in its own document, according to the Test Documentation Standard [IEEE829].
Test conditions are documented in a Test Design Specification. We will look at how to choose test conditions and prioritize them.
Test cases are documented in a Test Case Specification. We will look at how to write a good test case, showing clear traceability to the test basis (e.g. the requirement specification) as well as to test conditions.
Test procedures are documented (as you may expect) in a Test Procedure Specification (also known as a test script or a manual test script). We will look at how to translate test cases into test procedures relevant to the knowledge of the tester who will be executing the test, and we will look at how to produce a test execution schedule, using prioritization and technical and logical dependencies.
In this section, look for the definitions of the glossary terms: test case, test case specification, test condition, test data, test procedure specification, test script and traceability.
4.1.2 Formality of test documentation
Testing may be performed with varying degrees of formality. Very formal testing would have extensive documentation which is well controlled, and would expect the documented detail of the tests to include the exact and specific input and expected outcome of the test. Very informal testing may have no documenta-tion at all, or only notes kept by individual testers, but we'd still expect the testers to have in their minds and notes some idea of what they intended to test and what they expected the outcome to be. Most people are probably some-where in between! The right level of formality for you depends on your context: a commercial safety-critical application has very different needs than a one-off application to be used by only a few people for a short time.
The level of formality is also influenced by your organization - its culture, the people working there, how mature the development process is, how mature the testing process is, etc. The thoroughness of your test documentation may also depend on your time constraints; under excessive deadline pressure, keeping good documentation may be compromised.
In this chapter we will describe a fairly formal documentation style. If this is not appropriate for you, you might adopt a less formal approach, but you will be aware of how to increase formality if you need to.
4.1.3 Test analysis: identifying test conditions
Test analysis is the process of looking at something that can be used to derive test information. This basis for the tests is called the 'test basis'. It could be a system requirement, a technical specification, the code itself (for structural testing), or a business process. Sometimes tests can be based on an experienced user's knowledge of the system, which may not be documented. The test basis includes whatever the tests are based on. This was also discussed in Chapter 1. From a testing perspective, we look at the test basis in order to see what could be tested - these are the test conditions. A test condition is simply something that we could test. If we are looking to measure coverage of code decisions (branches), then the test basis would be the code itself, and the list of test conditions would be the decision outcomes (True and False). If we have a requirements specification, the table of contents can be our initial list of test conditions.
A good way to understand requirements better is to try to define tests to meet those requirements, as pointed out by [Hetzel, 1988].
For example, if we are testing a customer management and marketing system for a mobile phone company, we might have test conditions that are related to a marketing campaign, such as age of customer (pre-teen, teenager, young adult, mature), gender, postcode or zip code, and purchasing preference (pay-as-you-go or contract). A particular advertising campaign could be aimed at male teenaged customers in the mid-west of the USA on pay-as-you-go, for example.
Testing experts use different names to represent the basic idea of 'a list of things that we could test'. For example, Marick refers to 'test requirements' as things that should be tested. Although it is not intended to imply that everything must be tested, it is too easily interpreted in that way. [Marick, 1994] In con-trast, Hutcheson talks about the 'test inventory' as a list of things that could be tested [Hutcheson, 2003]; Craig talks about 'test objectives' as broad categories of things to test and 'test inventories' as the actual list of things that need to be tested [Craig, 2002]. These authors are all referring to what the ISTQB glossary calls a test condition.
When identifying test conditions, we want to 'throw the net wide' to identify as many as we can, and then we will start being selective about which ones to take forward to develop in more detail and combine into test cases. We could call them 'test possibilities'.
In Chapter 1 we explained that testing everything is known as exhaustive testing (defined as exercising every combination of inputs and preconditions) and we demonstrated that this is an impractical goal. Therefore, as we cannot test everything, we have to select a subset of all possible tests. In practice the subset we select may be a very small subset and yet it has to have a high proba-bility of finding most of the defects in a system. We need some intelligent thought processes to guide our selection; test techniques (i.e. test design tech-niques) are such thought processes.
A testing technique helps us select a good set of tests from the total number of all possible tests for a given system. Different techniques offer different ways of looking at the software under test, possibly challenging assumptions made about it. Each technique provides a set of rules or guidelines for the tester to follow in identifying test conditions and test cases. Techniques are described in detail later in this chapter.
The test conditions that are chosen will depend on the test strategy or detailed test approach. For example, they might be based on risk, models of the system, likely failures, compliance requirements, expert advice or heuristics. The word 'heuristic' comes from the same Greek root as eureka, which means 'I find'. A heuristic is a way of directing your attention, a common sense rule useful in solving a problem.
Test conditions should be able to be linked back to their sources in the test basis - this is called traceability.
Traceability can be either horizontal through all the test documentation for a given test level (e.g. system testing, from test conditions through test cases to test scripts) or vertical through the layers of development documentation (e.g. from requirements to components).
Why is traceability important? Consider these examples:
• The requirements for a given function or feature have changed. Some of the fields now have different ranges that can be entered. Which tests were looking at those boundaries? They now need to be changed. How many tests will actually be affected by this change in the requirements? These questions can be answered easily if the requirements can easily be traced to the tests.
• A set of tests that has run OK in the past has started to have serious prob lems. What functionality do these tests actually exercise? Traceability between the tests and the requirement being tested enables the functions or features affected to be identified more easily.
• Before delivering a new release, we want to know whether or not we have tested all of the specified requirements in the requirements specification. We have the list of the tests that have passed - was every requirement tested?
Having identified a list of test conditions, it is important to prioritize them, so that the most important test conditions are identified (before a lot of time is spent in designing test cases based on them). It is a good idea to try and think of twice as many test conditions as you need - then you can throw away the less important ones, and you will have a much better set of test conditions!
Note that spending some extra time now, while identifying test conditions, doesn't take very long, as we are only listing things that we could test. This is a good investment of our time - we don't want to spend time implementing poor tests!
Test conditions can be identified for test data as well as for test inputs and test outcomes, for example, different types of record, different distribution of types of record within a file or database, different sizes of records or fields in a record. The test data should be designed to represent the most important types of data, i.e. the most important data conditions.
Test conditions are documented in the IEEE 829 document called a Test Design Specification, shown below. (This document could have been called a Test Condition Specification, as the contents referred to in the standard are actually test conditions.)
IEEE 829 STANDARD:
TEST DESIGN SPECIFICATION TEMPLATE
Test design specification identifier Features to be tested Approach refinements Test identification Feature pass/fail criteria
4.1.4 Test design: specifying test cases
Test conditions can be rather vague, covering quite a large range of possibilities as we saw with our mobile phone company example (e.g. a teenager in the mid-west), or a test condition may be more specific (e.g. a particular male customer on pay-as-you-go with less than $10 credit). However when we come to make a test case, we are required to be very specific; in fact we now need exact and
detailed specific inputs, not general descriptions (e.g. Jim Green, age 17, living in Grand Rapids, Michigan, with credit of $8.64, expected result: add to Q4 marketing campaign). Note that one test case covers a number of conditions (teenager, male, mid-west area, pay-as-you-go, and credit of less than $10).
For a test condition of 'an existing customer', the test case input needs to be 'Jim Green' where Jim Green already exists on the customer database, or part of this test would be to set up a database record for Jim Green.
A test case needs to have input values, of course, but just having some values to input to the system is not a test! If you don't know what the system is sup-posed to do with the inputs, you can't tell whether your test has passed or failed.
Should these detailed test cases be written down? They can be formally doc-umented, as we will describe below. However, it is possible to test without doc-umenting at the test-case level. If you give an experienced user acceptance tester with a strong business background a list of high-level test conditions, they could probably do a good job of testing. But if you gave the same list to a new starter who didn't know the system at all, they would probably be lost, so they would benefit from having more detailed test cases.
Test cases can be documented as described in the IEEE 829 Standard for Test Documentation. Note that the contents described in the standard don't all have to be separate physical documents. But the standard's list of what needs to be kept track of is a good starting point, even if the test conditions and test cases for a given functionality or feature are all kept in one physical document.
One of the most important aspects of a test is that it assesses that the system does what it is supposed to do. Copeland says 'At its core, testing is the process of comparing "what is" with "what ought to be" '. [Copeland, 2003] If we simply put in some inputs and think 'that was fun, I guess the system is probably OK because it didn't crash', then are we actually testing it? We don't think so. You have observed that the system does what the system does - this is not a test. Boris Beizer refers to this as 'kiddie testing' [Beizer, 1990]. We may not know what the right answer is in detail every time, and we can still get some benefit from this approach at times, but it isn't really testing.
In order to know what the system should do, we need to have a source of informa-tion about the correct behavior of the system - this is called an 'oracle' or a test oracle. This has nothing to do with databases or companies that make them. It comes from the ancient Greek Oracle at Delphi, who supposedly could predict the future with unerring accuracy. Actually her answers were so vague that people interpreted them in whatever way they wanted - perhaps a bit like requirements specifications!
Once a given input value has been chosen, the tester needs to determine what the expected result of entering that input would be and document it as part of the test case.
Expected results include information displayed on a screen in response to an input, but they also include changes to data and/or states, and any other conse-quences of the test (e.g. a letter to be printed overnight).
What if we don't decide on the expected results before we run a test? We can still look at what the system produces and would probably notice if something was wildly wrong. However, we would probably not notice small differences in calculations, or results that seemed to look OK (i.e. are plausible). So we would conclude that the test had passed, when in fact the software has not given the correct result. Small differences in one calculation can add up to something very major later on, for example if results are multiplied by a large factor.
Ideally expected results should be predicted before the test is run - then your assessment of whether or not the software did the right thing will be more objective.
For a few applications it may not be possible to predict or know exactly what an expected result should be; we can only do a 'reasonableness check'. In this case we have a 'partial oracle' - we know when something is very wrong, but would probably have to accept something that looked reasonable. An example is when a system has been written to calculate something where it may not be possible to manually produce expected results in a reasonable timescale because the calculations are so complex.
In addition to the expected results, the test case also specifies the environ-ment and other things that must be in place before the test can be run (the pre-conditions) and any things that should apply after the test completes (the postconditions).
IEEE 829 STANDARD: j TEST CASE SPECIFICATION TEMPLATE [
Test case specification identifier Output specifications
Test items Environmental needs
Input specifications Special procedural requirements
Intercase dependencies
The test case should also say why it exists - i.e. the objective of the test it is part of or the test conditions that it is exercising (traceability). Test cases can now be prioritized so that the most important test cases are executed first, and low priority test cases are executed later, or even not executed at all. This may reflect the priorities already established for test conditions or the priority may be determined by other factors related to the specific test cases, such as a spe-cific input value that has proved troublesome in the past, the risk associated with the test, or the most sensible sequence of running the tests. Chapter 5 gives more detail of risk-based testing.
Test cases need to be detailed so that we can accurately check the results and know that we have exactly the right response from the system. If tests are to be automated, the testing tool needs to know exactly what to compare the system output to.
4.1.5 Test implementation: specifying test procedures or scripts
The next step is to group the test cases in a sensible way for executing them and to specify the sequential steps that need to be done to run the test. For example, a set of simple tests that cover the breadth of the system may form a regression suite, or all of the tests that explore the working of a given functionality or feature in depth may be grouped to be run together.
Some test cases may need to be run in a particular sequence. For example, a test may create a new customer record, amend that newly created record and
then delete it. These tests need to be run in the correct order, or they won't test what they are meant to test.
The document that describes the steps to be taken in running a set of tests (and specifies the executable order of the tests) is called a test proce-dure in IEEE 829, and is often also referred to as a test script. It could be called a manual test script for tests that are intended to be run manually rather than using a test execution tool. Test script is also used to describe the instructions to a test execution tool. An automation script is written in a programming language that the tool can interpret. (This is an automated test procedure.) See Chapter 6 for more information on this and other types of testing tools.
The test procedures, or test scripts, are then formed into a test execution schedule that specifies which procedures are to be run first - a kind of super-script. The test schedule would say when a given script should be run and by whom. The schedule could vary depending on newly perceived risks affecting the priority of a script that addresses that risk, for example. The logical and technical dependencies between the scripts would also be taken into account when scheduling the scripts. For example, a regression script may always be the first to be run when a new release of the software arrives, as a smoke test or sanity check.
Returning to our example of the mobile phone company's marketing cam-paign, we may have some tests to set up customers of different types on the database. It may be sensible to run all of the setup for a group of tests first. So our first test procedure would entail setting up a number of customers, includ-ing Jim Green, on the database.
Test procedure DB15: Set up customers for marketing campaign Y. Step 1: Open database with write privilege Step 2: Set up customer Bob Flounders
male, 62, Hudsonville, contract Step 3: Set up customer Jim Green
male, 17, Grand Rapids, pay-as-you-go, $8.64 Step 4: ...
We may then have another test procedure to do with the marketing cam-paign:
Test procedure MC03: Special offers for low-credit teenagers Step 1: Get details for Jim Green from database Step 2: Send text message offering double credit Step 3: Jim Green requests $20 credit, $40 credited
Writing the test procedure is another opportunity to prioritize the tests, to ensure that the best testing is done in the time available. A good rule of thumb is 'Find the scary stuff first'. However the definition of what is 'scary' depends on the business, system or project. For example, is it worse to raise Bob Founders' credit limit when he is not a good credit risk (he may not pay for the credit he asked for) or to refuse to raise his credit limit when he is a good credit risk (he may go elsewhere for his phone service and we lose the opportunity of lots of income from him).
IEEE 829 STANDARD:
TEST PROCEDURE SPECIFICATION TEMPLATE
Test procedure specification identifier
Purpose
Special requirements
Procedure steps
4.2 CATEGORIES OF TEST DESIGN TECHNIQUES
1 Recall reasons that both specification-based (black-box) and structure- based (white-box) approaches to test case design are useful, and list the common techniques for each. (Kl)
2 Explain the characteristics and differences between specification-based testing, structure-based testing and experience-based testing. (K2)
In this section we will look at the different types of test design technique, how they are used and how they differ. The three types or categories are distin-guished by their primary source: a specification, the structure of the system or component, or a person's experience. All categories are useful and the three are complementary.
In this section, look for the definitions of the glossary terms: black-box test design techniques, experience-based test design techniques, specification-based test design techniques, structure-based test design techniques and white-box test design techniques.
4.2.1 Introduction
There are many different types of software testing technique, each with its own strengths and weaknesses. Each individual technique is good at finding partic-ular types of defect and relatively poor at finding other types. For example, a technique that explores the upper and lower limits of a single input range is more likely to find boundary value defects than defects associated with combi-nations of inputs. Similarly, testing performed at different stages in the software development life cycle will find different types of defects; component testing is more likely to find coding logic defects than system design defects.
Each testing technique falls into one of a number of different categories. Broadly speaking there are two main categories, static and dynamic. Static tech-niques were discussed in Chapter 3. Dynamic techniques are subdivided into three more categories: specification-based (black-box, also known as behavioral
techniques), structure-based (white-box or structural techniques) and experi-ence-based. Specification-based techniques include both functional and non-functional techniques (i.e. quality characteristics). The techniques covered in the syllabus are summarized in Figure 4.1.
4.2.2 Static testing techniques
As we saw in Chapter 3, static testing techniques do not execute the code being examined and are generally used before any tests are executed on the software. They could be called non-execution techniques. Most static testing techniques can be used to 'test' any form of document including source code, design docu-ments and models, functional specifications and requirement specifications. However, 'static analysis' is a tool-supported type of static testing that concen-trates on testing formal languages and so is most often used to statically test source code.
4.2.3 Specification-based (black-box) testing techniques
The first of the dynamic testing techniques we will look at are the specification-based testing techniques. These are also known as 'black-box' or input/output-driven testing techniques because they view the software as a black-box with inputs and outputs, but they have no knowledge of how the system or
component is structured inside the box. In essence, the tester is concentrating on what the software does, not how it does it.
Notice that the definition mentions both functional and non-functional testing. Functional testing is concerned with what the system does, its features or functions. Non-functional testing is concerned with examining how well the system does something, rather than what it does. Non-functional aspects (also known as quality characteristics or quality attributes) include performance, usability, portability, maintainability, etc. Techniques to test these non-func-tional aspects are less procedural and less formalized than those of other cate-gories as the actual tests are more dependent on the type of system, what it does and the resources available for the tests.
Non-functional testing is part of the Syllabus and is also covered in Chapter 2. There are techniques for deriving non-functional tests [Gilb, 1988], [Testing Standards], but they are not covered at the Foundation level.
Categorizing tests into black and white-box is mentioned in a number of testing books, including [Beizer, 1990] and [Copeland, 2003].
4.2.4 Structure-based (white-box) testing techniques
Structure-based testing techniques (which are also dynamic rather than static) use the internal structure of the software to derive test cases. They are com-monly called 'white-box' or 'glass-box' techniques (implying you can see into the system) since they require knowledge of how the software is implemented, that is, how it works. For example, a structural technique may be concerned with exercising loops in the software. Different test cases may be derived to exercise the loop once, twice, and many times. This may be done regardless of the func-tionality of the software.
4.2.5 Experience-based testing techniques
In experience-based techniques, people's knowledge, skills and background are a prime contributor to the test conditions and test cases. The experience of both technical and business people is important, as they bring different perspectives to the test analysis and design process. Due to previous experience with similar systems, they may have insights into what could go wrong, which is very useful for testing.
4.2.6 Where to apply the different categories of techniques
Specification-based techniques are appropriate at all levels of testing (compo-nent testing through to acceptance testing) where a specification exists. When performing system or acceptance testing, the requirements specification or functional specification may form the basis of the tests. When performing com-ponent or integration testing, a design document or low-level specification forms the basis of the tests.
Structure-based techniques can also be used at all levels of testing. Developers use structure-based techniques in component testing and compo-nent integration testing, especially where there is good tool support for code coverage. Structure-based techniques are also used in system and acceptance
testing, but the structures are different. For example, the coverage of menu options or major business transactions could be the structural element in system or acceptance testing.
Experience-based techniques are used to complement specification-based and structure-based techniques, and are also used when there is no specifica-tion, or if the specification is inadequate or out of date. This may be the only type of technique used for low-risk systems, but this approach may be particu-larly useful under extreme time pressure - in fact this is one of the factors leading to exploratory testing.
4.3 SPECIFICATION-BASED OR BLACK-BOX TECHNIQUES
1 Write test cases from given software models using the following test design techniques. (K3)
a equivalence partitioning;
b boundary value analysis;
c decision tables;
d state transition testing.
2 Understand the main purpose of each of the four techniques, what level and type of testing could use the technique, and how coverage may be measured. (K2)
3 Understand the concept of use case testing and its benefits. (K2)
In this section we will look in detail at four specification-based or black-box techniques. These four techniques are K3 in the Syllabus - this means that you need to be able to use these techniques to design test cases. We will also cover briefly (not at K3 level) the specification-based technique of use case testing. In Section 4.4, we will look at the K3 structure-based techniques.
In this section, look for the definitions of the glossary terms: boundary value analysis, decision table testing, equivalence partitioning, state transition testing and use case testing.
The four specification-based techniques we will cover in detail are:
• equivalence partitioning;
• boundary value analysis;
• decision tables;
• state transition testing.
Note that we will discuss the first two together, because they are closely related.
4.3.1 Equivalence partitioning and boundary value analysis
Equivalence partitioning
Equivalence partitioning (EP) is a good all-round specification-based black-box technique. It can be applied at any level of testing and is often a good tech-nique to use first. It is a common sense approach to testing, so much so that most testers practise it informally even though they may not realize it. However, while it is better to use the technique informally than not at all, it is much better to use the technique in a formal way to attain the full benefits that it can deliver. This technique will be found in most testing books, including [Myers, 1979] and [Copeland, 2003].
The idea behind the technique is to divide (i.e. to partition) a set of test con-ditions into groups or sets that can be considered the same (i.e. the system should handle them equivalently), hence 'equivalence partitioning'. Equivalence partitions are also known as equivalence classes - the two terms mean exactly the same thing.
The equivalence-partitioning technique then requires that we need test only one condition from each partition. This is because we are assuming that all the conditions in one partition will be treated in the same way by the software. If one condition in a partition works, we assume all of the conditions in that par-tition will work, and so there is little point in testing any of these others. Conversely, if one of the conditions in a partition does not work, then we assume that none of the conditions in that partition will work so again there is little point in testing any more in that partition. Of course these are simplifying assumptions that may not always be right but if we write them down, at least it gives other people the chance to challenge the assumptions we have made and hopefully help to identify better partitions. If you have time, you may want to try more than one value from a partition, especially if you want to confirm a selection of typical user inputs.
For example, a savings account in a bank earns a different rate of interest depending on the balance in the account. In order to test the software that cal-culates the interest due, we can identify the ranges of balance values that earn the different rates of interest. For example, if a balance in the range $0 up to $100 has a 3% interest rate, a balance over $100 and up to $1000 has a 5% inter-est rate, and balances of $1000 and over have a 7% interest rate, we would ini-tially identify three valid equivalence partitions and one invalid partition as shown below.
Invalid partition Valid (for 3% interest) Valid (for 5%) Valid (for 7%)
-$0.01 $0.00 $100.00 $100.01 $999.99 $1000.00
Notice that we have identified four partitions here, even though the specifica-tion only mentions three. This illustrates a very important task of the tester - not only do we test what is in our specification, but we also think about things that haven't been specified. In this case we have thought of the situation where the balance is less than zero. We haven't (yet) identified an invalid partition on the right, but this would also be a good thing to consider. In order to identify where the 7% partition ends, we would need to know what the maximum balance is for this account (which may not be easy to find out). In our example we have left this open for the time being. Note that non-numeric input is also an invalid partition (e.g. the letter 'a') but we discuss only the numeric partitions for now.
We have made an assumption here about what the smallest difference is between two values. We have assumed two decimal places, i.e. $100.00, but we could have assumed zero decimal places (i.e. $100) or more than two decimal places (e.g. $100.0000) In any case it is a good idea to state your assumptions -then other people can see them and let you know if they are correct or not.
When designing the test cases for this software we would ensure that the three valid equivalence partitions are each covered once, and we would also test the invalid partition at least once. So for example, we might choose to calculate the interest on balances of-$10.00, $50.00, $260.00 and $1348.00. If we hadn't specifically identified these partitions, it is possible that at least one of them could have been missed at the expense of testing another one several times over. Note that we could also apply equivalence partitioning to outputs as well. In this case we have three interest rates: 3%, 5% and 7%, plus the error message for the invalid partition (or partitions). In this example, the output par-titions line up exactly with the input partitions.
How would someone test this without thinking about the partitions? A naive tester (let's call him Robbie) might have thought that a good set of tests would be to test every $50. That would give the following tests: $50.00, $100.00, $150.00, $200.00, $250.00, ... say up to $800.00 (then Robbie would have got tired of it and thought that enough tests had been carried out). But look at what Robbie has tested: only two out of four partitions! So if the system does not cor-rectly handle a negative balance or a balance of $1000 or more, he would not have found these defects - so the naive approach is less effective than equiva-lence partitioning. At the same time, Robbie has four times more tests (16 tests versus our four tests using equivalence partitions), so he is also much less effi-cient! This is why we say that using techniques such as this makes testing both more effective and more efficient.
Note that when we say a partition is 'invalid', it doesn't mean that it repre-sents a value that cannot be entered by a user or a value that the user isn't sup-posed to enter. It just means that it is not one of the expected inputs for this particular field. The software should correctly handle values from the invalid partition, by replying with an error message such as 'Balance must be at least $0.00'.
Note also that the invalid partition may be invalid only in the context of crediting interest payments. An account that is overdrawn will require some different action.
Boundary value analysis
Boundary value analysis (BVA) is based on testing at the boundaries between partitions. If you have ever done 'range checking', you were probably using the boundary value analysis technique, even if you weren't aware of it. Note that we have both valid boundaries (in the valid partitions) and invalid boundaries (in the invalid partitions).
To apply boundary value analysis, we will take the minimum and maximum (boundary) values from the valid partition (1 and 99 in this case) together with
As an example, consider a printer that has an input option of the number of copies to be made, from 1 to 99.
the first or last value respectively in each of the invalid partitions adjacent to the valid partition (0 and 100 in this case). In this example we would have three equivalence partitioning tests (one from each of the three partitions) and four boundary value tests.
Consider the bank system described in the section about equivalence parti-tioning.
Because the boundary values are defined as those values on the edge of a partition, we have identified the following boundary values: -$0.01 (an invalid boundary value because it is at the edge of an invalid partition), $0.00, $100.00, $100.01, $999.99 and $1000.00, all valid boundary values.
So by applying boundary value analysis we will have six tests for boundary values. Compare what our naive tester Robbie had done: he did actually hit one of the boundary values ($100) though it was more by accident than design. So in addition to testing only half of the partitions, Robbie has only tested one-sixth of the boundaries (so he will be less effective at finding any boundary defects). If we consider all of our tests for both equivalence partitioning and boundary value analysis, the techniques give us a total of nine tests, compared to the 16 that Robbie had, so we are still considerably more efficient as well as being over three times more effective (testing four partitions and six bound-aries, so 10 conditions in total compared to three).
Note that in the bank interest example, we have valid partitions next to other valid partitions. If we were to consider an invalid boundary for the 3% interest rate, we have -$0.01, but what about the value just above $100.00? The value of $100.01 is not an invalid boundary; it is actually a valid boundary because it falls into a valid partition. So the partition for 5%, for example, has no invalid boundary values associated with partitions next to it.
A good way to represent the valid and invalid partitions and boundaries is in a table such as Table 4.1:
TABLE 4.1 Equivalence partitions and boundaries
Test conditions Valid partitions Invalid partitions Valid boundaries Invalid boundaries
Balance in $0.00 $100.00 < $0.00 $0.00 -$0.01
aU°Unt $100.01-$999.99 >$Max $100.00 $Max+0.01
$1000.00- $Max non-integer (if $100.01
balance is an input field) $999.99
$1000.00 $Max
Interest rates 3% Any other value Not applicable Not applicable
5% Non-integer
7% No interest calculated
By showing the values in the table, we can see that no maximum has been specified for the 7% interest rate. We would now want to know what the maximum value is for an account balance, so that we can test that boundary. This is called an 'open boundary', because one of the sides of the partition is left open, i.e. not defined. But that doesn't mean we can ignore it - we should still try to test it, but how?
Open boundaries are more difficult to test, but there are ways to approach them. Actually the best solution to the problem is to find out what the bound-ary should be specified as! One approach is to go back to the specification to see if a maximum has been stated somewhere else for a balance amount. If so, then we know what our boundary value is. Another approach might be to investigate other related areas of the system. For example, the field that holds the account balance figure may be only six figures plus two decimal figures. This would give a maximum account balance of $999 999.99 so we could use that as our maximum boundary value. If we really cannot find any-thing about what this boundary should be, then we probably need to use an intuitive or experience-based approach to probe various large values trying to make it fail.
We could also try to find out about the lower open boundary - what is the lowest negative balance? Although we have omitted this from our example, setting it out in the table shows that we have omitted it, so helps us be more thorough if we wanted to be.
Representing the partitions and boundaries in a table such as this also makes it easier to see whether or not you have tested each one (if that is your objective).
Extending equivalence partitioning and boundary value analysis So far, by using EP and BVA we have identified conditions that could be tested, i.e. partitions and boundary values. The techniques are used to identify test con-ditions, which could be at a fairly high level (e.g. 'low-interest account') or at a detailed level (e.g. 'value of $100.00'). We have been looking at applying these techniques to ranges of numbers. However, we can also apply the techniques to other things.
For example, if you are booking a flight, you have a choice of Economy/Coach, Premium Economy, Business or First Class tickets. Each of these is an equivalence partition in its own right and should be tested, but it doesn't make sense to talk about boundaries for this type of partition, which is a collection of valid things. The invalid partition would be an attempt to type in any other type of flight class (e.g. 'Staff'). If this field is implemented using a drop-down list, then it should not be possible to type anything else in, but it is still a good test to try at least once in some drop-down field. When you are ana-lyzing the test basis (e.g. a requirements specification), equivalence partitioning can help to identify where a drop-down list would be appropriate.
When trying to identify a defect, you might try several values in a partition. If this results in different behavior where you expected it to be the same, then there may be two (or more) partitions where you initially thought there was only one.
We can apply equivalence partitioning and boundary value analysis to all levels of testing. The examples here were at a fairly detailed level probably most appropriate in component testing or in the detailed testing of a single screen.
At a system level, for example, we may have three basic configurations which our users can choose from when setting up their systems, with a number of options for each configuration. The basic configurations could be system administrator, manager and customer liaison. These represent three equiva-lence partitions that could be tested. We could have serious problems if we forget to test the configuration for the system administrator, for example.
We can also apply equivalence partitioning and boundary value analysis more than once to the same specification item. For example, if an internal tele-phone system for a company with 200 telephones has 3-digit extension numbers from 100 to 699, we can identify the following partitions and boundaries:
• digits (characters 0 to 9) with the invalid partition containing non-digits
• number of digits, 3 (so invalid boundary values of 2 digits and 4 digits)
• range of extension numbers, 100 to 699 (so invalid boundary values of 099 and 700)
• extensions that are in use and those that are not (two valid partitions, no boundaries)
• the lowest and highest extension numbers that are in use could also be used as boundary values
One test case could test more than one of these partitions/boundaries. For example, Extension 409 which is in use would test four valid partitions: digits, the number of digits, the valid range, and the 'in use' partition. It also tests the boundary values for digits, 0 and 9.
How many test cases would we need to test all of these partitions and boundaries, both valid and invalid? We would need a non-digit, a 2-digit and 4-digit number, the values of 99, 100, 699 and 700, one extension that is not in use, and possibly the lowest and highest extensions in use. This is ten or eleven test cases - the exact number would depend on what we could combine in one test case.
Compare this with the one-digit number example in Section 1.1.6. Here we found that we needed 68 tests just to test a one-digit field, if we were to test it thoroughly. Using equivalence partitioning and boundary value analysis helps us to identify tests that are most likely to find defects, and to use fewer test cases to find them. This is because the contents of a partition are representative of all of the possible values. Rather than test all ten individual digits, we test one in the middle (e.g. 4) and the two edges (0 and 9). Instead of testing every possi-ble non-digit character, one can represent all of them. So we have four tests (instead of 68) for a one-digit field.
As we mentioned earlier, we can also apply these techniques to output par-titions. Consider the following extension to our bank interest rate example. Suppose that a customer with more than one account can have an extra 1% interest on this account if they have at least $1000 in it. Now we have two pos-sible output values (7% interest and 8% interest) for the same account balance, so we have identified another test condition (8% interest rate). (We may also have identified that same output condition by looking at customers with more than one account, which is a partition of types of customer.)
Equivalence partitioning can be applied to different types of input as well. Our examples have concentrated on inputs that would be typed in by a (human) user when using the system. However, systems receive input data from other
sources as well, such as from other systems via some interface - this is also a good place to look for partitions (and boundaries). For example, the value of an interface parameter may fall into valid and invalid equivalence partitions. This type of defect is often difficult to find in testing once the interfaces have been joined together, so is particularly useful to apply in integration testing (either component integration or system integration).
Boundary value analysis can be applied to the whole of a string of charac-ters (e.g. a name or address). The number of characters in the string is a par-tition, e.g. between 1 and 30 characters is the valid partition with valid boundaries of 1 and 30. The invalid boundaries would be 0 characters (null, just hit the Return key) and 31 characters. Both of these should produce an error message.
Partitions can also be identified when setting up test data. If there are differ-ent types of record, your testing will be more representative if you include a data record of each type. The size of a record is also a partition with boundaries, so we could include maximum and minimum size records in the test database.
If you have some inside knowledge about how the data is physically organ-ized, you may be able to identify some hidden boundaries. For example, if an overflow storage block is used when more than 255 characters are entered into a field, the boundary value tests would include 255 and 256 characters in that field. This may be verging on white-box testing, since we have some knowledge of how the data is structured, but it doesn't matter how we classify things as long as our testing is effective at finding defects. Don't get hung up on a fine distinc-tion - just do whatever testing makes sense, based on what you know. An old Chinese proverb says, 'It doesn't matter whether the cat is white or black; all that matters is that the cat catches mice'.
With boundary value analysis, we think of the boundary as a dividing line between two things. Hence we have a value on each side of the boundary (but the boundary itself is not a value).
Invalid Valid Invalid
0" 1 99" TOO
Looking at the values for our printer example, 0 is in an invalid partition, 1 and 99 are in the valid partition and 100 is in the other invalid partition. So the boundary is between the values of 0 and 1, and between the values of 99 and 100. There is a school of thought that regards an actual value as a boundary value. By tradition, these are the values in the valid partition (i.e. the values specified). This approach then requires three values for every boundary two, so you would have 0,1 and 2 for the left boundary, and 98, 99 and 100 for the right boundary in this example. The boundary values are said to be 'on and either side of the boundary' and the value that is 'on' the boundary is generally taken to be in the valid partition.
Note that Beizer talks about domain testing, a generalization of equiva-lence partitioning, with three-value boundaries. He makes a distinction between open and closed boundaries, where a closed boundary is one where the point is included in the domain. So the convention is for the valid parti-tion to have closed boundaries. You may be pleased to know that you don't have to know this for the exam! British Standard 7925-2 Standard for Software Component Testing also defines a three-value approach to boundary value analysis.
So which approach is best? If you use the two-value approach together with equivalence partitioning, you are equally effective and slightly more efficient than the three-value approach. (We won't go into the details here but this can be demonstrated.) In this book we will use the two-value approach. In the exam, you may have a question based on either the two-value or the three-value approach, but it should be clear what the correct choice is in either case.
Designing test cases
Having identified the conditions that you wish to test, in this case by using equivalence partitioning and boundary value analysis, the next step is to design the test cases. The more test conditions that can be covered in a single test case, the fewer test cases will be needed in order to cover all the conditions. This is usually the best approach to take for positive tests and for tests that you are rea-sonably confident will pass. However if a test fails, then we need to find out why it failed - which test condition was handled incorrectly? We need to get a good balance between covering too many and too few test conditions in our tests.
Let's look at how one test case can cover one or more test conditions. Using the bank balance example, our first test could be of a new customer with a balance of $500. This would cover a balance in the partition from $100.01 to $999.99 and an output partition of a 5% interest rate. We would also be cover-ing other partitions that we haven't discussed yet. for example a valid customer, a new customer, a customer with only one account, etc. All of the partitions covered in this test are valid partitions.
When we come to test invalid partitions, the safest option is probably to try to cover only one invalid test condition per test case. This is because programs may stop processing input as soon as they encounter the first problem. So if you have an invalid customer name, invalid address, and invalid balance, you may get an error message saying 'invalid input' and you don"t know whether the test has detected only one invalid input or all of them. (This is also why specific error messages are much better than general ones!)
However, if it is known that the software under test is required to process all input regardless of its validity, then it is sensible to continue as before and design test cases that cover as many invalid conditions in one go as possible. For example, if every invalid field in a form has some red text above or below the field saying that this field is invalid and why, then you know that each field has been checked, so you have tested all of the error processing in one test case. In either case, there should be separate test cases covering valid and invalid con-ditions.
To cover the boundary test cases, it may be possible to combine all of the minimum valid boundaries for a group of fields into one test case and also the maximum boundary values. The invalid boundaries could be tested together if the validation is done on every field; otherwise they should be tested separately, as with the invalid partitions.
Why do both equivalence partitioning and boundary value analysis? Technically, because every boundary is in some partition, if you did only bound-ary value analysis you would also have tested every equivalence partition. However, this approach may cause problems if that value fails - was it only the boundary value that failed or did the whole partition fail? Also by testing only
boundaries we would probably not give the users much confidence as we are using extreme values rather than normal values. The boundaries may be more difficult (and therefore more costly) to set up as well.
For example, in the printer copies example described earlier we identified the following boundary values:
Suppose we test only the valid boundary values 1 and 99 and nothing in between. If both tests pass, this seems to indicate that all the values in between should also work. However, suppose that one page prints correctly, but 99 pages do not. Now we don't know whether any set of more than one page works, so the first thing we would do would be to test for say 10 pages, i.e. a value from the equivalence partition.
We recommend that you test the partitions separately from boundaries - this means choosing partition values that are NOT boundary values.
However, if you use the three-value boundary value approach, then you would have valid boundary values of 1, 2, 98 and 99, so having a separate equiv-alence value in addition to the extra two boundary values would not give much additional benefit. But notice that one equivalence value, e.g. 10, replaces both of the extra two boundary values (2 and 98). This is why equivalence partition-ing with two-value boundary value analysis is more efficient than three-value boundary value analysis.
Which partitions and boundaries you decide to exercise (you don't need to test them all), and which ones you decide to test first, depends on your test objectives. If your goal is the most thorough approach, then follow the pro-cedure of testing valid partitions first, then invalid partitions, then valid boundaries and finally invalid boundaries. However if you are under time pressure and cannot test everything (and who isn't?), then your test objec-tives will help you decide what to test. If you are after user confidence of typical transactions with a minimum number of tests, you may do valid par-titions only. If you want to find as many defects as possible as quickly as pos-sible, you may start with boundary values, both valid and invalid. If you want confidence that the system will handle bad inputs correctly, you may do mainly invalid partitions and boundaries. Your previous experience of types of defects found can help you find similar defects; for example if there are typically a number of boundary defects, then you would start by testing boundaries.
Equivalence partitioning and boundary value analysis are described in most testing books, including [Myers, 1979] and [Copeland, 2003]. Examples of types of equivalence classes to look out for are given in [Kaner et al., 1993] Equivalence partitioning and boundary value analysis are described in BS7925-2, including designing tests and measuring coverage.
4.3.2 Decision table testing
Why use decision tables?.
The techniques of equivalence partitioning and boundary value analysis are
often applied to specific situations or inputs. However, if different combinations
of inputs result in different actions being taken, this can be more difficult to show using equivalence partitioning and boundary value analysis, which tend to be more focused on the user interface. The other two specification-based tech-niques, decision tables and state transition testing are more focused on business logic or business rules.
A decision table is a good way to deal with combinations of things (e.g. inputs). This technique is sometimes also referred to as a 'cause-effect' table. The reason for this is that there is an associated logic diagramming technique called 'cause-effect graphing' which was sometimes used to help derive the decision table (Myers describes this as a combinatorial logic network [Myers, 1979]). However, most people find it more useful just to use the table described in [Copeland, 2003].
If you begin using decision tables to explore what the business rules are that should be tested, you may find that the analysts and developers find the tables very helpful and want to begin using them too. Do encourage this, as it will make your job easier in the future. Decision tables provide a systematic way of stating complex business rules, which is useful for developers as well as for testers. Decision tables can be used in test design whether or not they are used in specifications, as they help testers explore the effects of combinations of dif-ferent inputs and other software states that must correctly implement business rules. Helping the developers do a better job can also lead to better relation-ships with them.
Testing combinations can be a challenge, as the number of combinations can often be huge. Testing all combinations may be impractical if not impos-sible. We have to be satisfied with testing just a small subset of combinations but making the choice of which combinations to test and which to leave out is not trivial. If you do not have a systematic way of selecting combinations, an arbitrary subset will be used and this may well result in an ineffective test effort.
Decision tables aid the systematic selection of effective test cases and can have the beneficial side-effect of finding problems and ambiguities in the spec-ification. It is a technique that works well in conjunction with equivalence par-titioning. The combination of conditions explored may be combinations of equivalence partitions.
In addition to decision tables, there are other techniques that deal with testing combinations of things: pairwise testing and orthogonal arrays. These are described in [Copeland, 2003]. Another source of techniques is [Pol et al., 2001]. Decision tables and cause-effect graphing are described in [BS7925-2], including designing tests and measuring coverage.
Using decision tables for test design
The first task is to identify a suitable function or subsystem that has a behavior which reacts according to a combination of inputs or events. The behavior of interest must not be too extensive (i.e. should not contain too many inputs) oth-erwise the number of combinations will become cumbersome and difficult to manage. It is better to deal with large numbers of conditions by dividing them into subsets and dealing with the subsets one at a time.
Once you have identified the aspects that need to be combined, then you put them into a table listing all the combinations of True and False for each of the aspects. Take an example of a loan application, where you can enter the amount
of the monthly repayment or the number of years you want to take to pay it back (the term of the loan). If you enter both, the system will make a compro-mise between the two if they conflict. The two conditions are the loan amount and the term, so we put them in a table (see Table 4.2).
TABLE 4.2 Empty decision table
Conditions Rule 1 Rule 2 Rule 3 Rule 4
Repayment amount has been entered
Term of loan has been entered
Next we will identify all of the combinations of True and False (see Table 4.3). With two conditions, each of which can be True or False, we will have four combinations (two to the power of the number of things to be combined). Note that if we have three things to combine, we will have eight combinations, with four things, there are 16, etc. This is why it is good to tackle small sets of com-binations at a time. In order to keep track of which combinations we have, we will alternate True and False on the bottom row, put two Trues and then two Falses on the row above the bottom row, etc., so the top row will have all Trues and then all Falses (and this principle applies to all such tables).
TABLE 4.3 Decision table with input combinations
Conditions Rule 1 Rule 2 Rule 3 Rule 4
Repayment amount has T T F F
teen entered
Term of loan has been T F T F
entered
The next step (at least for this example) is to identify the correct outcome for each combination (see Table 4.4). In this example, we can enter one or both of the two fields. Each combination is sometimes referred to as a rule.
TABLE 4.4 Decision table with combinations and outcomes
Conditions Rule 1 Rule 2 Rule 3 Rule 4
Repayment amount has T T F F
been entered
Term of loan has been T F T F
entered
Actions/Outcomes
Process loan amount Y Y
Process term Y Y
At this point, we may realize that we hadn't thought about what happens if the customer doesn't enter anything in either of the two fields. The table has highlighted a combination that was not mentioned in the specification for this example. We could assume that this combination should result in an error message, so we need to add another action (see Table 4.5). This high-lights the strength of this technique to discover omissions and ambiguities in specifications. It is not unusual for some combinations to be omitted from specifications; therefore this is also a valuable technique to use when review-ing the test basis.
TABLE 4.5 Decision table with additional outcomes
Conditions Rule 1 Rule 2 Rule 3 Rule 4
Repayment amount has T T F F
been entered
Term of loan has been T F T F
entered
Actions/Outcomes
Process loan amount Y Y
Process term Y Y
Error message Y
Suppose we change our example slightly, so that the customer is not allowed to enter both repayment and term. Now our table will change, because there should also be an error message if both are entered, so it will look like Table 4.6.
TABLE 4.6 Decision table with changed outcomes
Conditions Rule 1 Rule 2 Rule 3 Rule 4
Repayment amount has T T F F
been entered
Term of loan has been T F T F
entered
Actions/Outcomes
Process loan amount Y
Process term Y
Error message Y Y
You might notice now that there is only one 'Yes' in each column, i.e. our actions are mutually exclusive - only one action occurs for each combination of conditions. We could represent this in a different way by listing the actions in the cell of one row, as shown in Table 4.7. Note that if more than one action results from any of the combinations, then it would be better to show them as separate rows rather than combining them into one row.
TABLE 4.7 Decision table with outcomes in one row
Conditions Rule 1 Rule 2 Rule 3 Rule 4
Repayment amount has T T F F
been entered
Term of loan has been T F T F
entered
Actions/Outcomes
Result Error Process loan Process Error
message amount term message
The final step of this technique is to write test cases to exercise each of the four rules in our table.
In this example we started by identifying the input conditions and then iden-tifying the outcomes. However in practice it might work the other way around - we can see that there are a number of different outcomes, and have to work back to understand what combination of input conditions actually drive those outcomes. The technique works just as well doing it in this way, and may well be an iterative approach as you discover more about the rules that drive the system.
Credit card worked example
Let's look at another example. If you are a new customer opening a credit card account, you will get a 15% discount on all your purchases today. If you are an existing customer and you hold a loyalty card, you get a 10% discount. If you have a coupon, you can get 20% off today (but it can't be used with the 'new customer' discount). Discount amounts are added, if applicable. This is shown in Table 4.8.
TABLE 4.8 Decision table for credit card example
In Table 4.8, the conditions and actions are listed in the left hand column. All the other columns in the decision table each represent a separate rule, one for each combination of conditions. We may choose to test each rule/combination and if there are only a few this will usually be the case. However, if the number of rules/combinations is large we are more likely to sample them by selecting a rich subset for testing.
Note that we have put X for the discount for two of the columns (Rules 1 and 2) - this means that this combination should not occur. You cannot be both a new customer and already hold a loyalty card! There should be an error message stating this, but even if we don't know what that message should be, it will still make a good test.
We have made an assumption in Rule 3. Since the coupon has a greater discount than the new customer discount, we assume that the customer will choose 20% rather than 15%. We cannot add them, since the coupon cannot be used with the 'new customer' discount. The 20% action is an assumption on our part, and we should check that this assumption (and any other assumptions that we make) is correct, by asking the person who wrote the specification or the users.
For Rule 5, however, we can add the discounts, since both the coupon and the loyalty card discount should apply (at least that's our assumption).
Rules 4, 6 and 7 have only one type of discount and Rule 8 has no discount. so 0%.
If we are applying this technique thoroughly, we would have one test for each column or rule of our decision table. The advantage of doing this is that we may test a combination of things that otherwise we might not have tested and that could find a defect.
However, if we have a lot of combinations, it may not be possible or sensible to test every combination. If we are time-constrained, we may not have time to test all combinations. Don't just assume that all combinations need to be tested; it is better to prioritize and test the most important combinations. Having the full table enables us to see which combinations we decided to test and which not to test this time.
There may also be many different actions as a result of the combinations of conditions. In the example above we just had one: the discount to be applied. The decision table shows which actions apply to each combination of conditions.
In the example above all the conditions are binary, i.e. they have only two possible values: True or False (or, if you prefer Yes or No). Often it is the case that conditions are more complex, having potentially many possible values. Where this is the case the number of combinations is likely to be very large, so the combinations may only be sampled rather than exercising all of them.
4.3.3 State transition testing
State transition testing is used where some aspect of the system can be described in what is called a 'finite state machine'. This simply means that the system can be in a (finite) number of different states, and the transitions from one state to another are determined by the rules of the 'machine'. This is the model on which the system and the tests are based. Any system where you get a different output for the same input, depending on what has happened before, is a finite state system. A finite state system is often shown as a state diagram (see Figure 4.2).
For example, if you request to withdraw $100 from a bank ATM, you may be given cash. Later you may make exactly the same request but be refused the money (because your balance is insufficient). This later refusal is because the state of your bank account has changed from having sufficient funds to cover
the withdrawal to having insufficient funds. The transaction that caused your account to change its state was probably the earlier withdrawal. A state diagram can represent a model from the point of view of the system, the account or the customer.
Another example is a word processor. If a document is open, you are able to close it. If no document is open, then 'Close' is not available. After you choose 'Close' once, you cannot choose it again for the same document unless you open that document. A document thus has two states: open and closed.
A state transition model has four basic parts:
• the states that the software may occupy (open/closed or funded/insufficient funds);
• the transitions from one state to another (not all transitions are allowed);
• the events that cause a transition (closing a file or withdrawing money);
• the actions that result from a transition (an error message or being given your cash).
Note that in any given state, one event can cause only one action, but that the same event - from a different state - may cause a different action and a dif-ferent end state.
We will look first at test cases that execute valid state transitions.
Figure 4.2 shows an example of entering a Personal Identity Number (PIN) to a bank account. The states are shown as circles, the transitions as lines with arrows and the events as the text near the transitions. (We have not shown the actions explicitly on this diagram, but they would be a message to the customer saying things such as 'Please enter your PIN'.)
The state diagram shows seven states but only four possible events (Card inserted, Enter PIN, PIN OK and PIN not OK). We have not specified all of the possible transitions here - there would also be a time-out from 'wait for PIN' and from the three tries which would go back to the start state after the time had elapsed and would probably eject the card. There would also be a transi-tion from the 'eat card' state back to the start state. We have not specified all the possible events either - there would be a 'cancel' option from 'wait for PIN'
and from the three tries, which would also go back to the start state and eject the card. The 'access account' state would be the beginning of another state diagram showing the valid transactions that could now be performed on the account.
However this state diagram, even though it is incomplete, still gives us infor-mation on which to design some useful tests and to explain the state transition technique.
In deriving test cases, we may start with a typical scenario. A sensible first test case here would be the normal situation, where the correct PIN is entered the first time. To be more thorough, we may want to make sure that we cover every state (i.e. at least one test goes through each state) or we may want to cover every transition. A second test (to visit every state) would be to enter an incorrect PIN each time, so that the system eats the card. We still haven't tested every transition yet. In order to do that, we would want a test where the PIN was incorrect the first time but OK the second time, and another test where the PIN was correct on the third try. These tests are probably less important than the first two.
Note that a transition does not need to change to a different state (although all of the transitions shown above do go to a different state). So there could be a transition from 'access account' which just goes back to 'access account' for an action such as 'request balance'.
Test conditions can be derived from the state graph in various ways. Each state can be noted as a test condition, as can each transition. In the Syllabus, we need to be able to identify the coverage of a set of tests in terms of transitions.
Going beyond the level expected in the Syllabus, we can also consider tran-sition pairs and triples and so on. Coverage of all individual transitions is also known as 0-switch coverage, coverage of transition pairs is l-switch coverage, coverage of transition triples is 2-switch coverage, etc. Deriving test cases from the state transition model is a black-box approach. Measuring how much you have tested (covered) is getting close to a white-box perspective. However, state transition testing is regarded as a black-box technique.
One of the advantages of the state transition technique is that the model can be as detailed or as abstract as you need it to be. Where a part of the system is more important (that is, requires more testing) a greater depth of detail can be modeled. Where the system is less important (requires less testing), the model can use a single state to signify what would otherwise be a series of different states.
Testing for invalid transitions
Deriving tests only from a state graph (also known as a state chart) is very good
for seeing the valid transitions, but we may not easily see the negative tests,
where we try to generate invalid transitions. In order to see the total number of
combinations of states and transitions, both valid and invalid, a state table is
useful.
The state table lists all the states down one side of the table and all the events that cause transitions along the top (or vice versa). Each cell then represents a state-event pair. The content of each cell indicates which state the system will move to, when the corresponding event occurs while in the associated state. This will include possible erroneous events - events that are not expected to happen in certain states. These are negative test conditions.
Table 4.9 lists the states in the first column and the possible inputs across the top row. So, for example, if the system is in State 1, inserting a card will take it to State 2. If we are in State 2, and a valid PIN is entered, we go to State 6 to access the account. In State 2 if we enter an invalid PIN, we go to State 3. We have put a dash in the cells that should be impossible, i.e. they represent invalid transitions from that state.
We have put a question mark for two cells, where we enter either a valid or invalid PIN when we are accessing the account. Perhaps the system will take our PIN number as the amount of cash to withdraw? It might be a good test! Most of the other invalid cells would be physically impossible in this example. Invalid (negative) tests will attempt to generate invalid transitions, transitions that shouldn't be possible (but often make good tests when it turns out they are possible).
A more extensive description of state machines is found in [Marick, 1994]. State transition testing is also described in [Craig, 2002], [Copeland, 2003], [Beizer, 1990] and [Broekman, 2003]. State transition testing is described in BS7925-2, including designing tests and coverage measures.
4.3.4 Use case testing
Use case testing is a technique that helps us identify test cases that exercise the whole system on a transaction by transaction basis from start to finish. They are described by Ivar Jacobson in his book Object-Oriented Software Engineering: A Use Case Driven Approach [Jacobson, 1992].
A use case is a description of a particular use of the system by an actor (a user of the system). Each use case describes the interactions the actor has with the system in order to achieve a specific task (or, at least, produce something of value to the user). Actors are generally people but they may also be other systems. Use cases are a sequence of steps that describe the interactions between the actor and the system.
Use cases are defined in terms of the actor, not the system, describing what the actor does and what the actor sees rather than what inputs the system expects and what the system'outputs. They often use the language and terms of the business rather than technical terms, especially when the actor is a business
user. They serve as the foundation for developing test cases mostly at the system and acceptance testing levels.
Use cases can uncover integration defects, that is, defects caused by the incorrect interaction between different components. Used in this way, the actor may be something that the system interfaces to such as a communication link or sub-system.
Use cases describe the process flows through a system based on its most likely use. This makes the test cases derived from use cases particularly good for finding defects in the real-world use of the system (i.e. the defects that the users are most likely to come across when first using the system). Each use case usually has a mainstream (or most likely) scenario and sometimes additional alternative branches (covering, for example, special cases or exceptional condi-tions). Each use case must specify any preconditions that need to be met for the use case to work. Use cases must also specify postconditions that are observable results and a description of the final state of the system after the use case has been executed successfully.
The PIN example that we used for state transition testing could also be defined in terms of use cases, as shown in Figure 4.3. We show a success sce-nario and the extensions (which represent the ways in which the scenario could fail to be a success).
For use case testing, we would have a test of the success scenario and one tesi for each extension. In this example, we may give extension 4b a higher priority than 4a from a security point of view.
System requirements can also be specified as a set of use cases. This approach can make it easier to involve the users in the requirements gathering and definition process.
4.4 STRUCTURE-BASED OR WHITE-BOX TECHNIQUES
1 Describe the concept and importance of code coverage. (K2)
2 Explain the concepts of statement and decision coverage and under stand that these concepts can also be used at other test levels than com ponent testing (e.g. on business procedures at system level). (K2)
3 Write test cases from given control flows using the following test design techniques: (K3)
a statement coverage; b decision coverage.
4 Assess statement and decision coverage for completeness. (K3)
In this section we will look in detail at the concept of coverage and how it can be used to measure some aspects of the thoroughness of testing. In order to see how coverage actually works, we will use some code-level examples (although coverage also applies to other levels such as business procedures). In particular, we will show how to measure coverage of statements and decisions, and how to write test cases to extend coverage if it is not 100%. The same principles apply to coverage of system-level coverage items, for example menu items.
In this section, look for the definitions of the glossary terms: code coverage, decision coverage, statement coverage, structural testing, structure-based testing and white-box testing
4.4.1 Using structure-based techniques to measure coverage and design tests
Structure-based techniques serve two purposes: test coverage measurement and structural test case design. They are often used first to assess the amount of testing performed by tests derived from specification-based techniques, i.e. to assess coverage. They are then used to design additional tests with the aim of increasing the test coverage.
Structure-based test design techniques are a good way of generating additional test cases that are different from existing tests. They can help ensure more breadth of testing, in the sense that test cases that achieve 100% coverage in any measure will be exercising all parts of the software from the point of view of the items being covered.
What is test coverage?
Test coverage measures in some specific way the amount of testing performed by a set of tests (derived in some other way, e.g. using specification-based techniques). Wherever we can count things and can tell whether or not each of those things has been tested by some test, then we can measure coverage. The basic coverage measure is
Number of coverage items exercised
Coverage = --------------------------------------------------- x 100%
Total number of coverage items
where the 'coverage item' is whatever we have been able to count and see whether a test has exercised or used this item.
There is danger in using a coverage measure. 100% coverage does not mean 100% tested! Coverage techniques measure only one dimension of a multi-dimen-sional concept. Two different test cases may achieve exactly the same coverage but the input data of one may find an error that the input data of the other doesn't.
One drawback of code coverage measurement is that it measures coverage of what has been written, i.e. the code itself; it cannot say anything about the software that has not been written. If a specified function has not been imple-mented, specification-based testing techniques will reveal this. If a function was omitted from the specification, then experience-based techniques may find it. But structure-based techniques can only look at a structure which is already there.
Types of coverage
Test coverage can be measured based on a number of different structural ele-ments in a system or component. Coverage can be measured at component-testing level, integration-testing level or at system- or acceptance-testing levels. For example, at system or acceptance level, the coverage items may be require-ments, menu options, screens, or typical business transactions. Other coverage measures include things such as database structural elements (records, fields and sub-fields) and files. It is worth checking for any new tools, as the test tool market develops quite rapidly.
At integration level, we could measure coverage of interfaces or specific interactions that have been tested. The call coverage of module, object or pro-cedure calls can also be measured (and is supported by tools to some extent).
We can measure coverage for each of the specification-based techniques as well:
• EP: percentage of equivalence partitions exercised (we could measure valid and invalid partition coverage separately if this makes sense);
• BVA: percentage of boundaries exercised (we could also separate valid and invalid boundaries if we wished);
• Decision tables: percentage of business rules or decision table columns tested;
• State transition testing: there are a number of possible coverage measures:
- Percentage of states visited
- Percentage of (valid) transitions exercised (this is known as Chow's 0- switch coverage)
- Percentage of pairs of valid transitions exercised ('transition pairs' or Chow's 1-switch coverage) - and longer series of transitions, such as tran sition triples, quadruples, etc.
- Percentage of invalid transitions exercised (from the state table)
The coverage measures for specification-based techniques would apply at whichever test level the technique has been used (e.g. system or component level).
When coverage is discussed by business analysts, system testers or users, it most likely refers to the percentage of requirements that have been tested by a set of tests. This may be measured by a tool such as a requirements manage-ment tool or a test management tool.
However, when coverage is discussed by programmers, it most likely refers to the coverage of code, where the structural elements can be identified using a tool, since there is good tool support for measuring code coverage. We will cover statement and decision coverage shortly.
Statements and decision outcomes are both structures that can be measured in code and there is good tool support for these coverage measures. Code cov-erage is normally done in component and component integration testing - if it is done at all. If someone claims to have achieved code coverage, it is important to establish exactly what elements of the code have been covered, as statement coverage (often what is meant) is significantly weaker than decision coverage or some of the other code-coverage measures.
How to measure coverage
For most practical purposes, coverage measurement is something that requires tool support. However, knowledge of the steps typically taken to measure cov-erage is useful in understanding the relative merits of each technique. Our example assumes an intrusive coverage measurement tool that alters the code by inserting instrumentation:
1 Decide on the structural element to be used, i.e. the coverage items to be counted.
2 Count the structural elements or items.
3 Instrument the code.
4 Run the tests for which coverage measurement is required.
5 Using the output from the instrumentation, determine the percentage of ele ments or items exercised.
Instrumenting the code (step 3) implies inserting code alongside each struc-tural element in order to record when that structural element has been exer-cised. Determining the actual coverage measure (step 5) is then a matter of analyzing the recorded information.
Coverage measurement of code is best done using tools (as described in Chapter 6) and there are a number of such tools on the market. These tools can help to increase quality and productivity of testing. They increase quality by ensuring that more structural aspects are tested, so defects on those structural paths can be found. They increase productivity and efficiency by highlighting tests that may be redundant, i.e. testing the same structure as other tests (although this is not necessarily a bad thing, since we may find a defect testing the same structure with different data).
In common with all structure-based testing techniques, code coverage tech-niques are best used on areas of software code where more thorough testing is required. Safety-critical code; code that is vital to the correct operation of a system, and complex pieces of code are all examples of where structure-based
techniques are particularly worth applying. For example, DO178-B [RTCA] requires structural coverage for certain types of system to be used by the mili-tary. Structural coverage techniques should always be used in addition to spec-ification-based and experience-based testing techniques rather than as an alternative to them.
Structure-based test case design
If you are aiming for a given level of coverage (say 95%) but you have not reached your target (e.g. you only have 87% so far), then additional test cases can be designed with the aim of exercising some or all of the struc-tural elements not yet reached. This is structure-based test design. These new tests are then run through the instrumented code and a new coverage measure is calculated. This is repeated until the required coverage measure is achieved (or until you decide that your goal was too ambi-tious!). Ideally all the tests ought to be run again on the un-instrumented code.
We will look at some examples of structure-based coverage and test design for statement and decision testing below.
4.4.2 Statement coverage and statement testing
Number of statements exercised
Statement coverage = --------------------------------------------- x 100%
Total number of statements
Statement coverage is calculated by:
Studies and experience in the industry have indicated that what is considered reasonably thorough black-box testing may actually achieve only 60% to 75% statement coverage. Typical ad hoc testing is likely to be around 30% - this leaves 70% of the statements untested.
Different coverage tools may work in slightly different ways, so they may give different coverage figures for the same set of tests on the same code, although at 100% coverage they should be the same.
We will illustrate the principles of coverage on code. In order to simplify our examples, we will use a basic pseudo-code - this is not any specific program-ming language, but should be readable and understandable to you, even if you have not done any programming yourself.
For example, consider code sample 4.1.
READ A
READB
IFA>BTHENC = 0
ENDIF
Code sample 4.1
To achieve 100% statement coverage of this code segment just one test case is required, one which ensures that variable A contains a value that is greater
than the value of variable B, for example, A = 12 and B = 10. Note that here we are doing structural test design first, since we are choosing our input values in order ensure statement coverage.
Let's look at an example where we measure coverage first. In order to sim-plify the example, we will regard each line as a statement. (Different tools and methods may count different things as statements, but the basic principle is the same however they are counted.) A statement may be on a single line, or it may be spread over several lines. One line may contain more than one statement, just one statement, or only part of a statement. Some statements can contain other statements inside them. In code sample 4.2, we have two read statements, one assignment statement, and then one IF statement on three lines, but the IF statement contains another statement (print) as part of it.
1 READ A
2 READ B
3 C =A + 2*B
4 IF C> 50 THEN
5 PRINT large C
6 ENDIF
Code sample 4.2
Although it isn't completely correct, we have numbered each line and will regard each line as a statement. (Some tools may group statements that would always be executed together in a basic block which is regarded as a single statement.) However, we will just use numbered lines to illustrate the principles of coverage of statements (lines). Let's analyze the coverage of a set of tests on our six-statement program:
TEST SET 1 Test 1_1: A = 2, B = 3 Test 1_2: A = 0, B = 25 Test 1_3: A = 47, B = 1
Which statements have we covered?
• In Test 1_1, the value of C will be 8, so we will cover the statements on lines 1 to 4 and line 6.
• In Test 1_2, the value of C will be 50, so we will cover exactly the same state ments as Test 1_1.
• In Test 1_3, the value of C will be 49, so again we will cover the same state ments.
Since we have covered five out of six statements, we have 83% statement coverage (with three tests). What test would we need in order to cover state-ment 5, the one statement that we haven't exercised yet? How about this one:
Test 1_4: A = 20, B = 25
This time the value of C is 70, so we will print 'Large C and we will have exercised all six of the statements, so now statement coverage = 100%. Notice that we measured coverage first, and then designed a test to cover the statement that we had not yet covered.
Note that Test 1_4 on its own is more effective- (towards our goal of achiev-ing 100% statement coverage) than the first three tests together. Just taking Test 1_4 on its own is also more efficient than the set of four tests, since it has used only one test instead of four. Being more effective and more efficient is the mark of a good test technique.
4.4.3 Decision coverage and decision testing
A decision is an IF statement, a loop control statement (e.g. DO-WHILE or REPEAT-UNTIL), or a CASE statement, where there are two or more possi-ble exits or outcomes from the statement. With an IF statement, the exit can either be TRUE or FALSE, depending on the value of the logical condition that comes after IF. With a loop control statement, the outcome is either to perform the code within the loop or not - again a True or False exit. Decision coverage is calculated by:
Number of decision outcomes exercised
Decision coverage = -------------------------------------------------------- x 100%
Total number of decision outcomes
What feels like reasonably thorough functional testing may achieve only 40% to 60% decision coverage. Typical ad hoc testing may cover only 20% of the decisions, leaving 80% of the possible outcomes untested. Even if your testing seems reasonably thorough from a functional or specification-based perspective, you may have only covered two-thirds or three-quarters of the decisions. Decision coverage is stronger than statement coverage. It 'sub-sumes' statement coverage - this means that 100% decision coverage always guarantees 100% statement coverage. Any stronger coverage measure may require more test cases to achieve 100% coverage. For example, consider code sample 4.1 again.
We saw earlier that just one test case was required to achieve 100% state-ment coverage. However, decision coverage requires each decision to have had both a True and False outcome. Therefore, to achieve 100% decision coverage, a second test case is necessary where A is less than or equal to B. This will ensure that the decision statement 'IF A > B' has a False outcome. So one test is sufficient for 100% statement coverage, but two tests are needed for 100% decision coverage. Note that 100% decision coverage guarantees 100% state-ment coverage, but not the other way around!
1 READ A
2 READ B
3 C=A-2*B
4 IFC <0THEN
5 PRINT "C negative"
6 ENDIF
Code sample 4.3
Let's suppose that we already have the following test, which gives us 100% statement coverage for code sample 4.3.
TEST SET 2
Test 2_1: A = 20, B = 15
Which decision outcomes have we exercised with our test? The value of C is -10, so the condition 'C < 0' is True, so we will print 'C negative' and we have exercised the True outcome from that decision statement. But we have not exer-cised the decision outcome of False. What other test would we need to exercise the False outcome and to achieve 100% decision coverage?
Before we answer that question, let's have a look at another way to represent this code. Sometimes the decision structure is easier to see in a control flow diagram (see Figure 4.4).
The dotted line shows where Test 2_1 has gone and clearly shows that we haven't yet had a test that takes the False exit from the IF statement.
Let's modify our existing test set by adding another test:
TEST SET 2
Test 2_1: A = 20, B = 15
Test 2_2: A = 10, B = 2
This now covers both of the decision outcomes, True (with Test 2_1) and False (with Test 2_2). If we were to draw the path taken by Test 2_2, it would be a straight line from the read statement down the False exit and through the ENDIF. Note that we could have chosen other numbers to achieve either the True or False outcomes.
4.4.4 Other structure-based techniques
There are other structure-based techniques that can be used to achieve testing to different degrees of thoroughness. Some techniques are stronger (require more tests to achieve 100% coverage and therefore, have a greater chance of detecting defects) and others are weaker.
For example, branch coverage is closely related to decision coverage and at 100% coverage they give exactly the same results. Decision coverage measures the coverage of conditional branches; branch coverage measures the coverage of both conditional and unconditional branches. The Syllabus uses decision
coverage, as it is the source of the branches. Some coverage measurement tools may talk about branch coverage when they actually mean decision coverage.
Other control-flow code-coverage measures include linear code sequence and jump (LCSAJ) coverage, condition coverage, multiple condition coverage (also known as condition combination coverage) and condition determination coverage (also known as multiple condition decision coverage or modified con-dition decision coverage, MCDC). This technique requires the coverage of all conditions that can affect or determine the decision outcome.
Another popular, but often misunderstood, code-coverage measure is path coverage. Sometimes any structure-based technique is called 'path testing' [Patton, 2001]. However, strictly speaking, for any code that contains a loop, path coverage is impossible since a path that travels round the loop three times is different from the path that travels round the same loop four times. This is true even if the rest of the paths are identical. So if it is possible to travel round the loop an unlimited number of times then there are an unlimited number of paths through that piece of code. For this reason it is more correct to talk about 'independent path segment coverage' though the shorter term 'path coverage is frequently used.
Structure-based measures and related test design techniques are described in [BS7925-2]. Structure-based techniques are also discussed in [Copeland. 2003] and [Myers, 1979]. A good description of the graph theory behind struc-tural testing can be found in [Jorgensen, 1995] and [Hetzel, 1988] also shows a structural approach. [Pol et al, 2001] describes a structure-based approach called an algorithm test.
4.5 EXPERIENCE-BASED TECHNIQUES
1 Recall reasons for writing test cases based on intuition, experience and knowledge about common defects. (Kl)
2 Compare experience-based techniques with specification-based testing techniques. (K2)
In this section we will look at two experience-based techniques, why and when they are useful, and how they fit with specification-based techniques.
Although it is true that testing should be rigorous, thorough and systematic, this is not all there is to testing. There is a definite role for non-systematic tech-niques, i.e. tests based on a person's knowledge, experience, imagination and intuition. The reason is that some defects are hard to find using more system-atic approaches, so a good 'bug hunter' can be very creative at finding those elusive defects.
In this section, look for the definitions of the glossary terms: error guessing and exploratory testing.
4.5.1 Error guessing
Error guessing is a technique that should always be used as a complement to other more formal techniques. The success of error guessing is very much dependent on the skill of the tester, as good testers know where the defects are most likely to lurk. Some people seem to be naturally good at testing and others are good testers because they have a lot of experience either as a tester or working with a particular system and so are able to pin-point its weaknesses. This is why an error-guessing approach, used after more formal techniques have been applied to some extent, can be very effective. In using more formal tech-niques, the tester is likely to gain a better understanding of the system, what it does and how it works. With this better understanding, he or she is likely to be better at guessing ways in which the system may not work properly.
There are no rules for error guessing. The tester is encouraged to think of situations in which the software may not be able to cope. Typical conditions to try include division by zero, blank (or no) input, empty files and the wrong kind of data (e.g. alphabetic characters where numeric are required). If anyone ever says of a system or the environment in which it is to operate 'That could never happen', it might be a good idea to test that condition, as such assumptions about what will and will not happen in the live environment are often the cause of failures. A structured approach to the error-guessing technique is to list pos-sible defects or failures and to design tests that attempt to produce them. These defect and failure lists can be built based on the tester's own experience or that of other people, available defect and failure data, and from common knowledge about why software fails.
4.5.2 Exploratory testing
Exploratory testing is a hands-on approach in which testers are involved in minimum planning and maximum test execution. The planning involves the cre-ation of a test charter, a short declaration of the scope of a short (1 to 2 hour) time-boxed test effort, the objectives and possible approaches to be used.
The test design and test execution activities are performed in parallel typi-cally without formally documenting the test conditions, test cases or test scripts. This does not mean that other, more formal testing techniques will not be used. For example, the tester may decide to use boundary value analysis but will think through and test the most important boundary values without necessarily writing them down. Some notes will be written during the exploratory-testing session, so that a report can be produced afterwards.
Test logging is undertaken as test execution is performed, documenting the key aspects of what is tested, any defects found and any thoughts about possible further testing. A key aspect of exploratory testing is learning: learning by the tester about the software, its use, its strengths and its weaknesses. As its name implies, exploratory testing is about exploring, finding out about the software, what it does, what it doesn't do, what works and what doesn't work. The tester is constantly making decisions about what to test next and where to spend the (limited) time.
This is an approach that is most useful when there are no or poor specifica-tions and when time is severely limited. It can also serve to complement other, more formal testing, helping to establish greater confidence in the software. In
this way, exploratory testing can be used as a check on the formal test process by helping to ensure that the most serious defects have been found.
Exploratory testing is described in [Kaner, 2002] and [Copeland, 2003] Other ways of testing in an exploratory way ('attacks') are described b\ [Whittaker, 2002].
4.6 CHOOSING A TEST TECHNIQUE
1 List the factors that influence the selection of the appropriate test design technique for a particular kind of problem, such as the type of system, risk, customer requirements, models for use case modeling, requirements models or tester knowledge. (K2)
In this final section we will look at the factors that go into the decision about which techniques to use when.
Which technique is best? This is the wrong question! Each technique is good for certain things, and not as good for other things. For example, one of the benefits of structure-based techniques is that they can find things in the code that aren't supposed to be there, such as 'Trojan horses' or other malicious code. However, if there are parts of the specification that are missing from the code, only specification-based techniques will find that -structure-based techniques can only test what is there. If there are things missing from the specification and from the code, then only experience-based techniques would find them. Each individual technique is aimed at particular types of defect as well. For example, state transition testing is unlikely to find boundary defects.
The choice of which test techniques to use depends on a number of factors, including the type of system, regulatory standards, customer or contractual requirements, level of risk, type of risk, test objective, documentation available, knowledge of the testers, time and budget, development life cycle, use case models and previous experience of types of defects found.
Some techniques are more applicable to certain situations and test levels: others are applicable to all test levels.
This chapter has covered the most popular and commonly used software testing techniques. There are many others that fall outside the scope of the Syllabus that this book is based on. With so many testing techniques to choose from how are testers to decide which ones to use?
Perhaps the single most important thing to understand is that the best testing technique is no single testing technique. Because each testing technique is good at finding one specific class of defect, using just one technique will help ensure that many (perhaps most but not all) defects of that particular class are found. Unfortunately, it may also help to ensure that many defects of other classes are missed! Using a variety of techniques will therefore help ensure that a variety of defects are found, resulting in more effective testing.
So how can we choose the most appropriate testing techniques to use? The decision will be based on a number of factors, both internal and external.
The internal factors that influence the decision about which technique to use are:
• Models used - Since testing techniques are based on models, the models available (i.e. developed and used during the specification, design and implementation of the system) will to some extent govern which testing techniques can be used. For example, if the specification contains a state transition diagram, state transition testing would be a good technique to use.
• Tester knowledge I experience - How much testers know about the system and about testing techniques will clearly influence their choice of testing tech niques. This knowledge will in itself be influenced by their experience of testing and of the system under test.
• Likely defects - Knowledge of the likely defects will be very helpful in choos ing testing techniques (since each technique is good at finding a particular type of defect). This knowledge could be gained through experience of testing a previous version of the system and previous levels of testing on the current version.
• Test objective - If the test objective is simply to gain confidence that the soft ware will cope with typical operational tasks then use cases would be a sen sible approach. If the objective is for very thorough testing then more rigorous and detailed techniques (including structure-based techniques) should be chosen.
• Documentation - Whether or not documentation (e.g. a requirements specification) exists and whether or not it is up to date will affect the choice of testing techniques. The content and style of the documentation will also influence the choice of techniques (for example, if decision tables or state graphs have been used then the associated test techniques should be used).
• Life cycle model - A sequential life cycle model will lend itself to the use of more formal techniques whereas an iterative life cycle model may be better suited to using an exploratory testing approach.
The external factors that influence the decision about which technique to use are:
• Risk - The greater the risk (e.g. safety-critical systems), the greater the need for more thorough and more formal testing. Commercial risk may be influ enced by quality issues (so more thorough testing would be appropriate) or by time-to-market issues (so exploratory testing would be a more appropri ate choice).
• Customer I contractual requirements - Sometimes contracts specify par ticular testing techniques to use (most commonly statement or branch coverage).
• Type of system - The type of system (e.g. embedded, graphical, financial, etc.) will influence the choice of techniques. For example, a financial application involving many calculations would benefit from boundary value analysis.
• Regulatory requirements - Some industries have regulatory standards or guidelines that govern the testing techniques used. For example, the aircraft industry requires the use of equivalence partitioning, boundary value analy sis and state transition testing for high integrity systems together with state ment, decision or modified condition decision coverage depending on the level of software integrity required.
• Time and budget - Ultimately how much time there is available will always affect the choice of testing techniques. When more time is available we can afford to select more techniques and when time is severely limited we will be limited to those that we know have a good chance of helping us find just the most important defects.
CHAPTER REVIEW
Let's review what you have learned in this chapter.
From Section 4.1, you should now be able to differentiate between a test condition, a test case and a test procedure, and know that they are documented in a test design specification, a test case specification and a test procedure specification respectively. You should be able to write test cases that include expected results and that show clear traceability to the test basis (e.g. requirements). You should be able to translate test cases into a test procedure specification at the appropriate level of detail for testers and you should be able to write a test execution schedule for a given set of test cases that takes into account priorities as well as technical and logical dependencies. You should know the glossary terms test cases, test case specification, test condition, test data, test procedure specification, test script and traceability.
From Section 4.2 (categories of test design techniques), you should be able to give reasons why both specification-based (black-box) and structure-based (white-box) approaches are useful, and list the common techniques for each of these approaches. You should be able to explain the characteristics and differences between specification-based, structure-based and experience-based techniques. You should know the glossary terms black-box test design techniques, experience-based test design techniques, specification-based test design techniques, structure-based test design techniques and white-box test design techniques.
From Section 4.3, you should be able to write test cases from given software models using equivalence partitioning, boundary value analysis, decision tables and state transition diagrams. You should understand the main purpose of each of these four techniques, what level and type of testing could use each tech-nique and how coverage can be measured for each of them. You should also understand the concept and benefits of use case testing. You should know the glossary terms boundary value analysis, decision table testing, equivalence partitioning, state transition testing and use case testing.
From Section 4.4, you should be able to describe the concept and importance of code coverage. You should be able to explain the concepts of statement and decision coverage and understand that these concepts can also be used at test levels other than component testing (such as business procedures at system test level). You should be able to write test cases from given control flows using statement testing and decision testing, and you should be able to assess statement and decision coverage for completeness. You should know the glossary terms code coverage, decision coverage, statement coverage, structural testing, structure-based testing and white-box testing.
From Section 4.5, you should be able to remember the reasons for writing test cases based on intuition, experience and knowledge about common defects and you should be able to compare experience-based techniques with specification-based techniques. You should know the glossary terms error guessing and exploratory testing.
From Section 4.6, you should be able to list the factors that influence the selection of the appropriate test design technique for a particular type of problem, such as the type of system, risk, customer requirements, models for use case modeling, requirements models or testing knowledge.
SAMPLE EXAM QUESTIONS
Question 1 In which document described in IEEE 829 would you find instructions for the steps to be taken for a test including set-up, logging, environment and measurement?
a. Test plan
b. Test design specification
c. Test case specification
d. Test procedure specification
Question 2 With a highly experienced tester with a good business background, which approach to defining test procedures would be effective and most efficient for a project under severe time pressure?
a. A high-level outline of the test conditions and general steps to take.
b. Every step in the test spelled out in detail.
c. A high-level outline of the test conditions with the steps to take discussed in detail with another experienced tester.
d. Detailed documentation of all test cases and careful records of each step taken in the testing.
Question 3 Put the test cases that implement the following test conditions into the best order for the test execution schedule, for a test that is checking modifications of customers on a database.
1 Print modified customer record.
2 Change customer address: house number and street name.
3 Capture and print the on-screen error message.
4 Change customer address: postal code.
5 Confirm existing customer is on the database by opening that record.
6 Close the customer record and close the database.
7 Try to add a new customer with no details at all.
a. 5,4, 2,1, 3, 7, 6
b. 4,2,5,1,6,7,3
c. 5,4,2,1,7,3,6
d. 5,1, 2, 3,4, 7, 6
Question 4 Why are both specification-based and structure-based testing techniques useful?
a. They find different types of defect.
b. Using more techniques is always better.
c. Both find the same types of defect.
d. Because specifications tend to be unstructured.
Question 5 What is a key characteristic of structure-based testing techniques?
a. They are mainly used to assess the structure of a specification.
b. They are used both to measure coverage and to design tests to increase coverage.
c. They are based on the skills and experience of the tester.
d. They use a formal or informal model of the software or component.
Question 6 Which of the following would be an example of decision-table testing for a financial application applied at the system-test level?
a. A table containing rules for combinations of inputs to two fields on a screen.
b. A table containing rules for interfaces between components.
c. A table containing rules for mortgage applications.
d. A table containing rules for chess.
Question 7 Which of the following could be a coverage measure for state transition testing?
V All states have been reached.
W The response time for each transaction is adequate.
X Every transition has been exercised.
Y All boundaries have been exercised.
Z Specific sequences of transitions have been exercised.
a. X,YandZ
b. V, X, Y and Z
c. W,XandY
d. V, X and Z
Question 8 Postal rates for 'light letters' are 25p up to l0g, 35p up to 50g plus an extra l0p for each additional 25g up to l00g.
Which test inputs (in grams) would be selected using equivalence partitioning?
a. 8,42,82,102
b. 4,15, 65, 92,159
c. 10,50,75,100
d. 5, 20, 40, 60, 80
Question 9 Which of the following could be used to assess the coverage achieved for specification-based (black-box) test techniques?
V Decision outcomes exercised W Partitions exercised
X Boundaries exercised
Y State transitions exercised Z Statements exercised
a. V,W,YorZ
b. W,XorY
c. V,XorZ
d. W, X, Y or Z
Question 10 Which of the following structure-based test design technique would be most likely to be applied to?
1 Boundaries between mortgage interest rate bands.
2 An invalid transition between two different arrears^ statuses.
3 The business process flow for mortgage approval.
4 Control flow of the program to calculate repayments.
a. 2, 3 and 4
b. 2 and 4
c. 3 and 4
d. 1, 2 and 3
Question 11 Use case testing is useful for which of the following?
P Designing acceptance tests with users or customers.
Q Making sure that the mainstream business processes are tested.
R Finding defects in the interaction between components.
S Identifying the maximum and minimum values
for every input field. T Identifying the percentage of statements
exercised by a sets of tests.
a. P, Q and R
b. Q,SandT
c. P, Qand S
d. R, S and T
Question 12 Which of the following statements about the relationship between statement coverage and decision coverage is correct?
a. 100% decision coverage is achieved if statement coverage is greater than 90%.
b. 100% statement coverage is achieved if decision coverage is greater than 90%.
c. 100% decision coverage always means 100% statement coverage.
d. 100% statement coverage always means 100% decision coverage.
Question 13 If you are flying with an economy ticket, there is a possibility that you may get upgraded to business class, especially if you hold a gold card in the airline's frequent flier program. If you don't hold a gold card, there is a possibility that you will get 'bumped' off the flight if it is full and you check in late. This is shown in Figure 4.5. Note that each box (i.e. statement) has been numbered.
Three tests have been run:
Test 1: Gold card holder who gets upgraded to business class
Test 2: Non-gold card holder who stays in economy Test 3: A person who is bumped from the flight What is the statement coverage of these three tests?
a. 60%
b. 70%
c. 80%
d. 90%
Question 14 Why are error guessing and exploratory testing good to do?
a. They can find defects missed by specification- based and structure-based techniques.
b. They don't require any training to be as effective as formal techniques.
c. They can be used most effectively when there are good specifications.
d. They will ensure that all of the code or system is tested.
Question 15 How do experience-based techniques differ from specification-based techniques?
a. They depend on the tester's understanding of the way the system is structured rather than on a documented record of what the system should do.
b. They depend on having older testers rather than younger testers.
c. They depend on a documented record of what the system should do rather than on an individual's personal view.
d. They depend on an individual's personal view rather than on a documented record of what the system should do.
Question 16 When choosing which technique to use in a given situation, which factors should be taken into account?
U previous experience of types of defects found in this or similar systems
V the existing knowledge of the testers
W regulatory standards that apply
X the type of test execution tool that will be used
Y the documentation available
Z previous experience in the development language
a. V,W,YandZ
b. U, V, W and Y
c. U,XandY
d. V, W and Y
Question 17 Given the state diagram in Figure 4.6, which test case is the minimum series of valid transitions to cover every state?
a. SS-S1-S2-S4-S1-S3-ES
b. SS-S1-S2-S3-S4-ES
c. SS-S1-S2-S4-S1-S3-S4-S1-S3-ES
d. SS-S1-S4-S2-S1-S3-ES
EXERCISES: TEST DESIGN TECHNIQUES
Exercises based on the techniques covered in this chapter are given in this section. Worked solutions are given in the next section.
Equivalence Partitioning/Boundary Value Analysis exercise
Scenario: If you take the train before 9:30 am or in the afternoon after 4:00 pm until 7:30 pm ('the rush hour'), you must pay full fare. A saver ticket is available for trains between 9:30 am and 4:00 pm, and after 7:30 pm.
What are the partitions and boundary values to test the train times for ticket types? Which are valid par-titions and which are invalid partitions? What are the boundary values? (A table may be helpful to organ-ize your partitions and boundaries.) Derive test cases for the partitions and boundaries.
Are there any questions you have about this 'requirement'? Is anything unclear?
Decision table exercise
Scenario: If you hold an 'over 60s' rail card, you get a 34% discount on whatever ticket you buy. If you are traveling with a child (under 16), you can get a 50% discount on any ticket if you hold a family rail card, otherwise you get a 10% discount. You can only hold one type of rail card.
Produce a decision table showing all the combinations of fare types and resulting discounts and derive test cases from the decision table.
State transition exercise
Scenario: A website shopping basket starts out as empty. As purchases are selected, they are added to the shopping basket. Items can also be removed from the shopping basket. When the customer decides to check out, a summary of the items in the basket and the total cost are shown, for the customer to say whether this is OK or not. If the contents and price are OK, then you leave the summary display and go to the payment system. Otherwise you go back to shopping (so you can remove items if you want).
a. Produce a state diagram showing the different states and transitions. Define a test, in terms of the sequence of states, to cover all transitions.
b. Produce a state table. Give an example test for an invalid transition.
Statement and decision testing exercise
Scenario: A vending machine dispenses either hot or cold drinks. If you choose a hot drink (e.g. tea or coffee), it asks if you want milk (and adds milk if required), then it asks if you want sugar (and adds sugar if required), then your drink is dispensed.
a. Draw a control flow diagram for this example. (Hint: regard the selection of the type of drink as one statement.)
b. Given the following tests, what is the statement coverage achieved? What is the decision coverage achieved?
Test 1: Cold drink
Test 2: Hot drink with milk and sugar
c. What additional tests would be needed to achieve 100% statement coverage? What additional tests would be needed to achieve 100% decision coverage?
EXERCISE SOLUTIONS
EP/BVA exercise
The first thing to do is to establish exactly what the boundaries are between the full fare and saver fare. Let's put these in a table to organize our thoughts:
We have assumed that the boundary values are: 9:29 am, 9:30 am, 4:00 pm, 4:01 pm, 7:30 pm and 7:31 pm. By setting out exactly what we think is meant by the specification, we may highlight some ambiguities or, at least, raise some questions - this is one of the benefits of using the technique! For example:
'When does the morning rush hour start? At midnight? At 11:30 pm the previous day? At the time of the first train of the day? If so, when is the first train? 5:00 am?'
This is a rather important omission from the specification. We could make an assumption about when it starts, but it would be better to find out what is correct.
• If a train is due to leave at exactly 4:00 pm, is a saver ticket still valid?
• What if a train is due to leave before 4:00 pm but is delayed until after 4:00 pm? Is a saver ticket still valid? (i.e. if the actual departure time is different to the scheduled departure time)
Our table above has helped us to see where the partitions are. All of the partitions in the table above are valid partitions. It may be that an invalid partition would be a time that no train was running, e.g. before 5:00 am, but our specification didn't mention that! However it would be good to show this possibility also. We could be a bit more formal by listing all valid and invalid partitions and boundaries in a table, as we described in Section 4.3.1, but in this case it doesn't actually add a lot, since all partitions are valid.
Here are the test cases we can derive for this example:
Note that test cases 1, 4, 7 and 10 are based on equivalence partition values; test cases 2, 3, 5, 6, 8 and 9 are based on boundary values. There may also be other information about the test cases, such as precondi-tions, that we have not shown here.
Decision table exercise
The fare types mentioned include an 'over 60s' rail card, a family rail card, and whether you are traveling with a child or not. With three conditions or causes, we have eight columns in our decision table below.
When we come to fill in the effects, we may find this a bit more difficult. For the first two rules, for example, what should the output be? Is it an X because holding more than one rail card should not be pos-sible? The specification doesn't actually say what happens if someone does hold more than one card, i.e. it has not specified the output, so perhaps we should put a question mark in this column. Of course, if someone does hold two rail cards, they probably wouldn't admit this, and perhaps they would claim the 50% discount with their family rail card if they are traveling with a child, so perhaps we should put 50% for Rule 1 and 34% for Rule 2 in this column. Our notation shows that we don't know what the expected outcome should be for these rules!
This highlights the fact that our natural language (English) specification is not very clear as to what the effects should actually be. A strength of this technique is that it forces greater clarity. If the answers are spelled out in a decision table, then it is clear what the effect should be. When different people come up with different answers for the outputs, then you have an unclear specification!
The word 'otherwise' in the specification is ambiguous. Does 'otherwise' mean that you always get at least a 10% discount or does it mean that if you travel with a child and an over 60s card but not a family card you get 10% and 34%? Depending on what assumption you make for the meaning of 'otherwise', you will get a different last row in your decision table.
Note that the effect or output is the same (34%) for both Rules 3 and 4. This means that our third cause (whether or not a child is also traveling) actually has no influence on the output. These columns could therefore be combined with 'don't care' as the entry for the third cause. This 'rationalizing' of the table means we will have fewer columns and therefore fewer test cases. The reduction in test cases is based on the assumption we are making about the factor having no effect on the outcome, so a more thorough approach would be to include each column in the table.
Here is a rationalized table, where we have shown our assumptions about the first two outcomes and we have also combined Rules 6 and 8 above, since having a family rail card has no effect if you are not travel-ing with a child.
Here are the test cases that we derive from this table. (If you didn't rationalize the table, then you will have eight test cases rather than six.) Note that you wouldn't necessarily test each column, but the table enables you to make a decision about which combinations to test and which not to test this time.
Note that we may have raised some additional issues when we designed the test cases. For example, does the discount for a rail card apply only to the traveler or to someone traveling with them? Here we have assumed that it applies to all travelers for the family rail card, but to the individual passenger only for the over 60s rail card.
State transition exercise
The state diagram is shown in Figure 4.7. The initial state (SI) is when the shopping basket is empty. When an item is added to the basket, it goes to state (S2), where there are potential purchases. Any additional items added to the basket do not change the state (just the total number of things to purchase). Items can be removed, which does not change the state unless the total items ordered goes from 1 to 0. In this case, we go back to the empty basket (SI). When we want to check out, we go to the summary state (S3) for approval. If the list and prices are approved, we go to payment (S4); if not, we go back to the shopping state (possibly to remove some items to reduce the total price we have to pay). There are four states and seven transitions.
Note that SI is our Start State for this example and S4 is the End State - this means that we are not con-cerned with any event that happens once we get to State S4.
Here is a test to cover all transitions. Note that the end state from one step or event is the start state for the next event, so these steps must be done in this sequence.
Although our example is not interested in what happens from State 4, there would be other events and actions once we enter the payment process that could be shown by another state diagram (e.g. check valid-ity of the credit card, deduct the amount, email a receipt, etc.).
The corresponding state table is:
All of the boxes that contain - are invalid transitions in this example. Example negative tests would include:
• attempt to add an item from the summary and cost state (S3)
• try to remove an item from the empty shopping basket (SI)
• try to enter 'OK' while in the Shopping state (S2).
Statement and decision testing exercise
The control flow diagram is shown in Figure 4.8. Note that drawing a control diagram here illustrates that structural testing can also be applied to the structure of general processes, not just to computer algorithms. Flowcharts are generally easier to understand than text when you are trying to describe the results of decisions taken on later events.
On Figure 4.9, we can see the route that Tests 1 and 2 have taken through our control flow graph. Test 1 has gone straight down the left-hand side to select a cold drink. Test 2 has gone to the right at each opportunity, adding both milk and sugar to a hot drink.
Every statement (represented by a box on the diagram) has been covered by our two tests, so we have 100% statement coverage.
We have not taken the No exit from either the 'milk?' or 'sugar?' decisions, so there are two decision outcomes that we have not tested yet. We did test both of the outcomes from the 'hot or cold?' decision, so we have covered four out of six decision outcomes. Decision coverage is 4/6 or 67% with the two tests.
No additional tests are needed to achieve statement coverage, as we already have 100% coverage of the statements.
One additional test is needed to achieve 100% decision coverage:
Test 3: Hot drink, no milk, no sugar
This test will cover both of the 'No' decision outcomes from the milk and sugar decisions, so we will now have 100% decision coverage.
CHAPTER 5 Test management
esting is a complex activity. Testing is often a distinct sub-project within the larger software development, maintenance, or integration project. Testing usually accounts for a substantial proportion of the overall project budget. Therefore, we must understand how we should manage the testing we do.
In this chapter, we cover essential topics for test management in six sections. The first relates to how to organize the testers and the testing. The second concerns the estimation, planning and strategizing of the test effort. The third addresses test progress monitoring, test reporting and test control. The fourth explains configuration management and its relationship to testing. The fifth covers the central topic of risk and how testing affects and is affected by product and project risks. The sixth and final section discusses the management of incidents, both product defects and other events that require further investigation.
5.1 TEST ORGANIZATION
1 Recognize the importance of independent testing. (Kl)
2 List the benefits and drawbacks of independent testing within an organ ization. (K2)
3 Recognize the different team members to be considered for the creation of a test team. (Kl)
4 Recall the tasks of typical test leaders and testers. (Kl)
In this section, let's talk about organizing a test effort within a project. We'll look at the value of independent testing, and discuss the potential benefits and risks associated with independent testing. We will examine the various types of different test team members we might want on a test team. And we'll familiarize ourselves with the typical tasks performed by test leaders and testers.
As we go through this section, keep your eyes open for the glossary terms tester, test leader and test manager.
5.1.1 Independent and integrated testing
In Chapter 1 we talked about independent testing from the perspective of indi-vidual tester psychology. In this chapter, we'll look at the organizational and managerial implications of independence.
The approaches to organizing a test team vary, as do the places in the organ-ization structure where the test team fits. Since testing is an assessment of quality, and since that assessment is not always positive, many organizations strive to create an organizational climate where testers can deliver an inde-pendent, objective assessment of quality.
When thinking about how independent the test team is, recognize that inde-pendence is not an either/or condition, but a continuum. At one end of the continuum lies the absence of independence, where the programmer performs testing within the programming team.
Moving toward independence, you find an integrated tester or group of testers working alongside the programmers, but still within and reporting to the development manager. You might find a team of testers who are independ-ent and outside the development team, but reporting to project management.
Near the other end of the continuum lies complete independence. You might see a separate test team reporting into the organization at a point equal to the development or project team. You might find specialists in the business domain (such as users of the system), specialists in technology (such as data-base experts), and specialists in testing (such as security testers, certification testers, or test automation experts) in a separate test team, as part of a larger independent test team, or as part of a contract, outsourced test team. Let's examine the potential benefits and risks of independence, starting with the benefits.
An independent tester can often see more, other, and different defects than a tester working within a programming team - or a tester who is by profession a programmer. While business analysts, marketing staff, designers, and pro-grammers bring their own assumptions to the specification and implementa-tion of the item under test, an independent tester brings a different set of assumptions to testing and to reviews, which often helps expose hidden defects and problems related to the group's way of thinking, as we discussed in Chapter 3. An independent tester brings a skeptical attitude of professional pessimism, a sense that, if there's any doubt about the observed behavior, they should ask: Is this a defect?
At the team level, an independent test team reporting to a senior or execu-tive manager may enjoy (once they earn it) more credibility in the organization than a test leader or tester who is part of the programming team. An independ-ent tester who reports to senior management can report his results honestly and without concern for reprisals that might result from pointing out problems in coworkers' or, worse yet, the manager's work. An independent test team often has a separate budget, which helps ensure the proper level of money is spent on tester training, testing tools, test equipment, and so forth. In addition, in some organizations, testers in an independent test team may find it easier to have a career path that leads up into more senior roles in testing.
Independent test teams are not risk-free. It's possible for the testers and the test team to become isolated. This can take the form of interpersonal isolation from the programmers, the designers, and the project team itself, or it can take
the form of isolation from the broader view of quality and the business objec-tives (e.g., obsessive focus on defects, often accompanied by a refusal to accept business prioritization of defects). This leads to communication problems, feelings of alienation and antipathy, a lack of identification with and support for the project goals, spontaneous blame festivals and political backstabbing.
Even well-integrated test teams can suffer problems. Other project stake-holders might come to see the independent test team - rightly or wrongly - as a bottleneck and a source of delay. Some programmers abdicate their respon-sibility for quality, saying, 'Well, we have this test team now, so why do I need to unit test my code?'
Due to a desire for the benefits of an independent test team, companies sometimes establish them, only to break them up again later. Why does that happen? A common cause is the failure of the test manager to effectively manage the risks of independence listed above. Some test teams succumb to the temptation to adopt a 'no can do' attitude, coming up with reasons why the project should bend to their needs rather than each side being flexible so as to enable project success. Testers take to acting as enforcers of process or as auditors without a proper management mandate and support. Resentments and pressures build, until at last the organization decides that the independent test team causes more problems than it solves. It's espe-cially important for testers and test managers to understand the mission they serve and the reasons why the organization wants an independent test team. Often, the entire test team must realize that, whether they are part of the project team or independent, they exist to provide a service to the project team.
There is no one right approach to organizing testing. For each project, you must consider whether to use an independent test team, based on the project, the application domain, and the levels of risk, among other factors. As the size, complexity, and criticality of the project increases, it is important to have inde-pendence in later levels of testing (like integration test, system test and accept-ance test), though some testing is often best done by other people such as project managers, quality managers, developers, business and domain experts or infrastructure or IT operations experts.
5.1.2 Working as a test leader
We have seen that the location of a test team within a project organization can vary widely. Similarly there is wide variation in the roles that people within the test team play. Some of these roles occur frequently, some infrequently. Two roles that are found within many test teams are those of the test leader and the tester, though the same people may play both roles at various points during the project. Let's take a look at the work done in these roles, starting with the test leader.
Test leaders tend to be involved in the planning, monitoring, and control of the testing activities and tasks discussed in Section 1.5 on the fundamental test process. At the outset of the project, test leaders, in collaboration with the other stakeholders, devise the test objectives, organizational test policies (if not already in place), test strategies and test plans. They estimate the testing to be done and negotiate with management to acquire the necessary resources. They recognize when test automation is appropriate and, if it is, they plan the
effort, select the tools, and ensure training of the team. They may consult with other groups - e.g., programmers - to help them with their testing. They lead, guide and monitor the analysis, design, implementation and execution of the test cases, test procedures and test suites. They ensure proper configuration management of the testware produced and traceability of the tests to the test basis.
As test execution comes near, they make sure the test environment is put into place before test execution and managed during test execution. They schedule the tests for execution and then they monitor, measure, control and report on the test progress, the product quality status and the test results, adapting the test plan and compensating as needed to adjust to evolving con-ditions. During test execution and as the project winds down, they write summary reports on test status.
Sometimes test leaders wear different titles, such as test manager or test coordinator. Alternatively, the test leader role may wind up assigned to a project manager, a development manager or a quality assurance manager. (Regarding the first two people on this list, warning bells about independence should be ringing in your head now, in addition to thoughts about how we can ensure that such non-testers gain the knowledge and outlook needed to manage testing.) Whoever is playing the role, expect them to plan, monitor and control the testing work.
5.1.3 Working as a tester
As with test leaders, projects should include testers at the outset, though it is often the case that project doesn't need a full complement of testers until the test execution period. In the planning and preparation phases of the testing, testers should review and contribute to test plans, as well as analyzing, review-ing and assessing requirements and design specifications. They may be involved in or even be the primary people identifying test conditions and cre-ating test designs, test cases, test procedure specifications and test data, and may automate or help to automate the tests. They often set up the test envi-ronments or assist system administration and network management staff in doing so.
As test execution begins, the number of testers often increases, starting with the work required to implement tests in the test environment. (They may play such a role on all test levels, even those not under the direct control of the test group; e.g., they might implement unit tests which were designed by program-mers.) Testers execute and log the tests, evaluate the results and document problems found. They monitor the testing and the test environment, often using tools for this task, and often gather performance metrics. Throughout the testing life cycle, they review each other's work, including test specifica-tions, defect reports and test results.
5.1.4 Defining the skills test staff need
Doing testing properly requires more than defining the right positions and number of people for those positions. Good test teams have the right mix of
skills based on the tasks and activities they need to carry out, and people outside the test team who are in charge of test tasks need the right skills, too. People involved in testing need basic professional and social qualifications such as literacy, the ability to prepare and deliver written and verbal reports, the ability to communicate effectively, and so on. Going beyond that, when we think of the skills that testers need, three main areas come to mind:
• Application or business domain: A tester must understand the intended behavior, the problem the system will solve, the process it will automate and so forth, in order to spot improper behavior while testing and recognize the 'must work' functions and features.
• Technology: A tester must be aware of issues, limitations and capabilities of the chosen implementation technology, in order to effectively and effi ciently locate problems and recognize the 'likely to fail' functions and features.
• Testing: A tester must know the testing topics discussed in this book - and often more advanced testing topics - in order to effectively and efficiently carry out the test tasks assigned.
The specific skills in each area and the level of skill required vary by project, organization, application, and the risks involved.
The set of testing tasks and activities are many and varied, and so too are the skills required, so we often see specialization of skills and separation of roles. For example, due to the special knowledge required in the areas of testing, technology and business domain, respectively, test tool experts may handle automating the regression tests, programmers may perform compo-nent and integration tests and users and operators may be involved in acceptance tests.
We have long advocated pervasive testing, the involvement of people throughout the project team in carrying out testing tasks. Let us close this section, though, on a cautionary note. Software and system companies (e.g., producers of shrink-wrapped software and consumer products) typi-cally overestimate the technology knowledge required to be an effective tester. Businesses that use information technology (e.g., banks and insur-ance companies) typically overestimate the business domain knowledge needed.
All types of projects tend to underestimate the testing knowledge required. We have seen a project fail in part because people without proper testing skills tested critical components, leading to the disastrous discovery of fundamental architectural problems later. Most projects can benefit from the participation of professional testers, as amateur testing alone will usually not suffice.
5.2 TEST PLANS, ESTIMATES AND STRATEGIES
1 Recognize the different levels and objectives of test planning. (Kl)
2 Summarize the purpose and content of the test plan, test design specifi cation and test procedure documents according to [IEEE 829]. (K2)
3 Recall typical factors that influence the effort related to testing. (Kl)
4 Differentiate between two conceptually different estimation approaches: the metrics-based approach and the expert-based approach. (K2)
5 Differentiate between the subject of test planning for a project, for indi vidual test levels (e.g. system test) or specific test targets (e.g. usability test), and for test execution. (K2)
6 List test preparation and execution tasks that need planning. (Kl)
7 Recognize and justify adequate exit criteria for specific test levels and groups of test cases (e.g. for integration testing, acceptance testing or usability testing). (K2)
In this section, let's talk about a complicated trio of test topics: plans, estimates and strategies. Plans, estimates and strategies depend on a number of factors, including the level, targets and objectives of the testing we're setting out to do. Writing a plan, preparing an estimate and selecting test strategies tend to happen concurrently and ideally during the planning period for the overall project, though we must ready to revise them as the project proceeds and we gain more information.
Let's look closely at how to prepare a test plan, examining issues related to planning for a project, for a test level or phase, for a specific test type and for test execution. We'll examine typical factors that influence the effort related to testing, and see two different estimation approaches: metrics-based and expert-based. We'll discuss selecting test strategies and ways to establish adequate exit criteria for testing. In addition, we'll look at various tasks related to test preparation and execution that need planning.
As you read, keep your eyes open for the glossary terms entry criteria, exit criteria, exploratory testing, test approach, test level, test plan, test procedure and test strategy.
5.2.1 The purpose and substance of test plans
While people tend to have different definitions of what goes in a test plan, for us a test plan is the project plan for the testing work to be done. It is not a test design specification, a collection of test cases or a set of test procedures; in fact, most of our test plans do not address that level of detail.
Why do we write test plans? We have three main reasons.
First, writing a test plan guides our thinking. We find that if we can explain something in words, we understand it. If not, there's a good chance we don't.
Writing a test plan forces us to confront the challenges that await us and focus our thinking on important topics. In Chapter 2 of Fred Brooks' brilliant and essential book on software engineering management, The Mythical Man-Month, he explains the importance of careful estimation and planning for testing as follows:
Failure to allow enough time for system test, in particular, is peculiarly disastrous. Since the delay comes at the end of the schedule, no one is aware of schedule trouble until almost the delivery date [and] delay at this point has unusually severe ... financial repercussions. The project is fully staffed, and cost-per-day is maximum [as are the associated opportunity costs]. It is therefore very important to allow enough system test time in the original schedule.
[Brooks, 1995]
We find that using a template when writing test plans helps us remember the important challenges. You can use the IEEE 829 test plan template shown in this chapter, use someone else's template, or create your own template over time.
The test planning process and the plan itself serve as vehicles for communi-cating with other members of the project team, testers, peers, managers and other stakeholders. This communication allows the test plan to influence the project team and the project team to influence the test plan, especially in the areas of organization-wide testing policies and motivations; test scope, objec-tives and critical areas to test; project and product risks, resource considera-tions and constraints; and the testability of the item under test.
You can accomplish this communication through circulation of one or two test plan drafts and through review meetings. Such a draft will include many notes such as the following:
[To Be Determined: Jennifer: Please tell me what the plan is for releasing the test items into the test lab for each cycle of system test execution?] [Dave - please let me know which version of the test tool will be used for the regression tests of the previous increments.]
As you document the answers to these kinds of questions, the test plan becomes a record of previous discussions and agreements between the testers and the rest of the project team.
The test plan also helps us manage change. During early phases of the project, as we gather more information, we revise our plans. As the project evolves and situations change, we adapt our plans. Written test plans give us a baseline against which to measure such revisions and changes. Furthermore, updating the plan at major milestones helps keep testing aligned with project needs. As we run the tests, we make final adjustments to our plans based on the results. You might not have the time - or the energy - to update your test plans every time a variance occurs, as some projects can be quite dynamic. In Chapter 6 [Black, 2001], we describe a simple approach for documenting variances from the test plan that you can implement using a database or spreadsheet. You can include these change records in a periodic test plan update, as part of a test status report, or as part as an end-of-project test summary.
We've found that it's better to write multiple test plans in some situations. For example, when we manage both integration and system test levels, those two test execution periods occur at different points in time and have different
objectives. For some systems projects, a hardware test plan and a software test plan will address different techniques and tools as well as different audiences. However, since there might be overlap between these test plans, a master test plan that addresses the common elements can reduce the amount of redundant documentation.
IEEE 829 STANDARD TEST PLAN TEMPLATE Test plan identifier Test deliverables
Introduction Test tasks
Test items Environmental needs
Features to be tested Responsibilities
Features not to be tested Staffing and training needs
Approach Schedule
Item pass/fail criteria Risks and contingencies
Suspension and resumption criteria Approvals
5.2.2 What to do with your brain while planning tests
Writing a good test plan is easier than writing a novel, but both tasks require an organized approach and careful thought. In fact, since a good test plan is kept short and focused, unlike some novels, some might argue that it's harder to write a good test plan. Let's look at some of the planning tasks you need to carry out.
At a high level, you need to consider the purpose served by the testing work. In terms of the overall organizational needs, this purpose is referred to vari-ously as the test team's mission or the organization's testing policy. In terms of the specific project, understanding the purpose of testing means knowing the answers to questions such as:
• What is in scope and what is out of scope for this testing effort?
• What are the test objectives?
• What are the important project and product risks? (more on risks in Section 5.5).
• What constraints affect testing (e.g., budget limitations, hard deadlines, etc.)?
• What is most critical for this product and project?
• Which aspects of the product are more (or less) testable?
• What should be the overall test execution schedule and how should we decide the order in which to run specific tests? (Product and planning risks, discussed later in this chapter, will influence the answers to these questions.)
You should then select strategies which are appropriate to the purpose of testing (more on the topic of selecting strategies in a few pages).
In addition, you need to decide how to split the testing work into various levels, as discussed in Chapter 2 (e.g., component, integration, system and acceptance). If that decision has already been made, you need to decide how to best fit your testing work in the level you are responsible for with the testing work done in those other test levels. During the analysis and design of tests, you'll want to reduce gaps and overlap between levels and, during test execu-tion, you'll want to coordinate between the levels. Such details dealing with inter-level coordination are often addressed in the master test plan.
In addition to integrating and coordinating between test levels, you should also plan to integrate and coordinate all the testing work to be done with the rest of the project. For example, what items must be acquired for the testing? Are there on-going supply issues, such as with imitation bills (i.e., simulated banknotes) for a financial application such as an ATM? When will the program-mers complete work on the system under test? What operations support is required for the test environment? What kind of information must be delivered to the maintenance team at the end of testing?
Moving down into the details, what makes a plan a plan - rather than a state-ment of principles, a laundry list of good ideas or a collection of suggestions -is that the author specifies in it who will do what when and (at least in a general way) how. Resources are required to carry out the work. These are often hard decisions that require careful consideration and building a consensus across the team, including with the project manager.
The entire testing process - from planning through to closure - produces information, some of which you will need to document. How precisely should testers write the test designs, cases and procedures? How much should they leave to the judgment of the tester during test execution, and what are the reproducibility issues associated with this decision? What kinds of templates can testers use for the various documents they'll produce? How do those docu-ments relate to one another? If you intend to use tools for tasks such as test design and execution, as discussed in Chapter 6, you'll need to understand how those tools capture documentation while working on the plan.
Some information you'll need to gather in the form of raw data and then distill. What metrics to do you intend to use to monitor, control and manage the testing? Which of those metrics - and perhaps other metrics - will you use to report your results? We'll look more closely at possible answers to those ques-tions in Section 5.3, but a good test plan provides answers early in the project.
Finally, moving back up to a higher level, think about what would be true about the project when the project was ready to start executing tests. What would be true about the project when the project was ready to declare test exe-cution done? At what point can you safely start a particular test level or phase, test suite or test target? When can you finish it? The factors to consider in such decisions are often called 'entry criteria' and 'exit criteria.' For such criteria, typical factors are:
• Acquisition and supply: the availability of staff, tools, systems and other materials required.
• Test items: the state that the items to be tested must be in to start and to finish testing.
• Defects: the number known to be present, the arrival rate, the number pre dicted to remain, and the number resolved.
• Tests: the number run, passed, failed, blocked, skipped, and so forth.
• Coverage: the portions of the test basis, the software code or both that have been tested and which have not.
• Quality: the status of the important quality characteristics for the system.
• Money: the cost of finding the next defect in the current level of testing com pared to the cost of finding it in the next level of testing (or in production).
• Risk: the undesirable outcomes that could result from shipping too early (such as latent defects or untested areas) - or too late (such as loss of market share).
When writing exit criteria, we try to remember that a successful project is a balance of quality, budget, schedule and feature considerations. This is even more important when applying exit criteria at the end of the project.
5.2.3 Estimating what testing will involve and what it will cost
The testing work to be done can often be seen as a subproject within the larger project. So, we can adapt fundamental techniques of estimation for testing. We could start with a work-breakdown structure that identifies the stages, activities and tasks.
Starting at the highest level, we can break down a testing project into phases using the fundamental test process identified in the ISTQB Syllabus: planning and control; analysis and design; implementation and execution; evaluating exit criteria and reporting; and test closure. Within each phase we identify activities and within each activity we identify tasks and perhaps subtasks. To identify the activities and tasks, we work both forward and backward. When we say we work forward, we mean that we start with the planning activities and then move forward in time step by step, asking, 'Now, what comes next?'
Working backward means that we consider the risks that we identified during risk analysis (which we'll discuss in Section 5.5). For those risks which you intend to address through testing, ask yourself, 'So, what activities and tasks are required in each stage to carry out this testing?' Let's look at an example of how you might work backward.
Suppose that you've identified performance as a major area of risk for your product. So, performance testing is an activity in the test execution phase. You now estimate the tasks involved with running a performance test, how long those tasks will take and how many times you'll need to run the performance tests.
Now, those tests didn't just appear out of thin air: someone had to develop them. So, performance test development entails activities in test analysis, design and implementation. You now estimate the tasks involved in developing a per-formance test, such as writing test scripts and creating test data.
Typically, performance tests need to be run in a special test environment that is designed to look like the production or field environment, at least in those ways which would affect response time and resource utilization and perform-ance tests need special tools to generate load and check response. So, perform-ance test environment acquisition and configuration is an activity in the test implementation phase. You now estimate tasks involved in acquiring and con-figuring such a test environment, such as simulating performance based on the
production environment design to look for potential bottlenecks, getting the right hardware, software and tools and setting up hardware, software and tools.
Not everyone knows how to use performance-testing tools or to design per-formance tests. So, performance-testing training or staffing is an activity in the test planning phase. Depending on the approach you intend to take, you now estimate the time required to identify and hire a performance test professional or to train one or more people in your organization to do the job.
Finally, in many cases a detailed test plan is written for performance testing, due to its differences from other test types. So, performance-testing planning is an activity in the test planning phase. You now estimate the time required to draft, review and finalize a performance test plan.
When you are creating your work-breakdown structure, remember that you will want to use it for both estimation (at the beginning) and monitoring and control (as the project continues). To ensure accuracy of the estimate and precise control, make sure that you subdivide the work finely enough. This means that tasks should be short in duration, say one to three days. If they are much longer - say two weeks - then you run the risk that long and complex sub-tasks are 'hiding' within the larger task, only to be discovered later. This can lead to nasty surprises during the project.
5.2.4 Estimation techniques
There are two techniques for estimation covered by the ISTQB Foundation Syllabus. One involves consulting the people who will do the work and other people with expertise on the tasks to be done. The other involves analyzing metrics from past projects and from industry data. Let's look at each in turn.
Asking the individual contributors and experts involves working with experi-enced staff members to develop a work-breakdown structure for the project. With that done, you work together to understand, for each task, the effort, duration, dependencies, and resource requirements. The idea is to draw on the collective wisdom of the team to create your test estimate. Using a tool such as Microsoft Project or a whiteboard and sticky-notes, you and the team can then predict the testing end-date and major milestones. This technique is often called 'bottom up' estimation because you start at the lowest level of the hier-archical breakdown in the work-breakdown structure - the task - and let the duration, effort, dependencies and resources for each task add up across all the tasks.
Analyzing metrics can be as simple or sophisticated as you make it. The sim-plest approach is to ask, 'How many testers do we typically have per developer on a project?' A somewhat more reliable approach involves classifying the project in terms of size (small, medium or large) and complexity (simple, mod-erate or complex) and then seeing on average how long projects of a particular size and complexity combination have taken in the past. Another simple and reliable approach we have used is to look at the average effort per test case in similar past projects and to use the estimated number of test cases to estimate the total effort. Sophisticated approaches involve building mathematical models in a spreadsheet that look at historical or industry averages for certain key parameters - number of tests run by tester per day, number of defects found by tester per day, etc. - and then plugging in those parameters to predict
duration and effort for key tasks or activities on your project. The tester-to-developer ratio is an example of a top-down estimation technique, in that the entire estimate is derived at the project level, while the parametric technique is bottom-up, at least when it is used to estimate individual tasks or activities.
We prefer to start by drawing on the team's wisdom to create the work-breakdown structure and a detailed bottom-up estimate. We then apply models and rules of thumb to check and adjust the estimate bottom-up and top-down using past history. This approach tends to create an estimate that is both more accurate and more defensible than either technique by itself.
Even the best estimate must be negotiated with management. Negotiating sessions exhibit amazing variety, depending on the people involved. However, there are some classic negotiating positions. It's not unusual for the test leader or manager to try to sell the management team on the value added by the testing or to alert management to the potential problems that would result from not testing enough. It's not unusual for management to look for smart ways to accelerate the schedule or to press for equivalent coverage in less time or with fewer resources. In between these positions, you and your colleagues can reach compromise, if the parties are willing. Our experience has been that successful negotiations about estimates are those where the focus is less on winning and losing and more about figuring out how best to balance competing pressures in the realms of quality, schedule, budget and features.
5.2.5 Factors affecting test effort
Testing is a complex endeavor on many projects and a variety of factors can influence it. When creating test plans and estimating the testing effort and schedule, you must keep these factors in mind or your plans and estimates will deceive you at the beginning of the project and betray you at the middle or end.
The test strategies or approaches you pick will have a major influence on the testing effort. This factor is so influential that we'll come back to it in Section 5.2.6. In this section, let's look at factors related to the product, the process and the results of testing.
Product factors start with the presence of sufficient project documentation so that the testers can figure out what the system is, how it is supposed to work and what correct behavior looks like. In other words, adequate and high-quality information about the test basis will help us do a better, more efficient job of defining the tests.
The importance of non-functional quality characteristics such as usability, reliability, security, performance, and so forth also influences the testing effort. These test targets can be expensive and time consuming.
Complexity is another major product factor. Examples of complexity consid-erations include:
• The difficulty of comprehending and correctly handling the problem the system is being built to solve (e.g., avionics and oil exploration software);
• The use of innovative technologies, especially those long on hyperbole and short on proven track records;
• The need for intricate and perhaps multiple test configurations, especially when these rely on the timely arrival of scarce software, hardware and other supplies;
• The prevalence of stringent security rules, strictly regimented processes or other regulations;
• The geographical distribution of the team, especially if the team crosses time-zones (as many outsourcing efforts do).
While good project documentation is a positive factor, it's also true that having to produce detailed documentation, such as meticulously specified test cases, results in delays. During test execution, having to maintain such detailed documentation requires lots of effort, as does working with fragile test data that must be maintained or restored frequently during testing.
Finally, increasing the size of the product leads to increases in the size of the project and the project team. Increases in the project and project team increases the difficulty of predicting and managing them. This leads to the dis-proportionate rate of collapse of large projects.
Process factors include the availability of test tools, especially those that reduce the effort associated with test execution, which is on the critical path for release. On the development side, debugging tools and a dedicated debugging environment (as opposed to debugging in the test environment) also reduce the time required to complete testing.
The life cycle itself is an influential process factor, as the V-model tends to be more fragile in the face of late change while incremental models tend to have high regression testing costs. Process maturity, including test process maturity, is another factor, especially the implication that mature processes involve carefully managing change in the middle and end of the project, which reduces test execution cost.
Time pressure is another factor to be considered. Pressure should not be an excuse to take unwarranted risks. However, it is a reason to make careful, con-sidered decisions and to plan and re-plan intelligently throughout the process, which is another hallmark of mature processes.
People execute the process, and people factors are as important or more important than any other. Indeed, even when many troubling things are true about a project, an excellent team can often make good things happen on the project and in testing. Important people factors include the skills of the individ-uals and the team as a whole, and the alignment of those skills with the project's needs. Since a project team is a team, solid relationships, reliable execution of agreed-upon commitments and responsibilities and a determination to work together towards a common goal are important. This is especially important for testing, where so much of what we test, use, and produce either comes from, relies upon or goes to people outside the testing group. Because of the impor-tance of trusting relationships and the lengthy learning curves involved in soft-ware and system engineering, the stability of the project team is an important people factor, too.
The test results themselves are important in the total amount of test effort during test execution. The delivery of good-quality software at the start of test execution and quick, solid defect fixes during test execution prevents delays in the test execution process. A defect, once identified, should not have to go through multiple cycles of fix/retest/re-open, at least not if the initial estimate is going to be held to.
You probably noticed from this list that we included a number of factors outside the scope and control of the test leader or manager. Indeed, events that
occur before or after testing can bring these factors about. For this reason, it's important that testers, especially test leaders or managers, be attuned to the overall context in which they operate. Some of these contextual factors result in specific project risks for testing, which should be addressed in the test plan. Project risks are discussed in more detail in Section 5.5.
5.2.6 Test approaches or strategies
The choice of test approaches or strategies is one powerful factor in the success of the test effort and the accuracy of the test plans and estimates. This factor is under the control of the testers and test leaders. Of course, having choices also means that you can make mistakes, so we'll talk about how to pick the right test strategies in a minute. First, though, let's survey the major types of test strate-gies that are commonly found.1
• Analytical: For example, the risk-based strategy involves performing a risk analysis using project documents and stakeholder input, then planning, esti mating, designing, and prioritizing the tests based on risk. (We'll talk more about risk analysis later in this chapter.) Another analytical test strategy is the requirements-based strategy, where an analysis of the requirements spec ification forms the basis for planning, estimating and designing tests. Analytical test strategies have in common the use of some formal or infor mal analytical technique, usually during the requirements and design stages of the project.
• Model-based: For example, you can build mathematical models for loading and response for e-commerce servers, and test based on that model. If the behavior of the system under test conforms to that predicted by the model, the system is deemed to be working. Model-based test strategies have in common the creation or selection of some formal or informal model for critical system behaviors, usually during the requirements and design stages of the project.
• Methodical: For example, you might have a checklist that you have put together over the years that suggests the major areas of testing to run or you might follow an industry-standard for software quality, such as ISO 9126, for your outline of major test areas. You then methodically design, implement and execute tests following this outline. Methodical test strategies have in common the adherence to a pre-planned, systematized approach that has been developed in-house, assembled from various concepts developed in- house and gathered from outside, or adapted significantly from outside ideas and may have an early or late point of involvement for testing.
• Process- or standard-compliant: For example, you might adopt the IEEE 829 standard for your testing, using books such as [Craig, 2002] or [Drabick, 2004] to fill in the methodological gaps. Alternatively, you might adopt one of the agile methodologies such as Extreme Programming. Process- or stan dard-compliant strategies have in common reliance upon an externally developed approach to testing, often with little - if any - customization and may have an early or late point of involvement for testing.
1 The catalog of testing strategies that has been included in the ISTQB Foundation Syllabus grew out of an email discussion between Rex Black, Ross Collard, Kathy Iberle and Cem Kaner. We thank them for their thought-provoking comments.
• Dynamic: For example, you might create a lightweight set of testing guide lines that focus on rapid adaptation or known weaknesses in software. Dynamic strategies, such as exploratory testing, have in common concentrat ing on finding as many defects as possible during test execution and adapt ing to the realities of the system under test as it is when delivered, and they typically emphasize the later stages of testing. See, for example, the attack- based approach of [Whittaker, 2002] and [Whittaker, 2003] and the exploratory approach of [Kaner et al., 2002].
• Consultative or directed: For example, you might ask the users or develop ers of the system to tell you what to test or even rely on them to do the testing. Consultative or directed strategies have in common the reliance on a group of non-testers to guide or perform the testing effort and typically emphasize the later stages of testing simply due to the lack of recognition of the value of early testing.
• Regression-averse: For example, you might try to automate all the tests of system functionality so that, whenever anything changes, you can re-run every test to ensure nothing has broken. Regression-averse strategies have in common a set of procedures - usually automated - that allow them to detect regression defects. A regression-averse strategy may involve automating functional tests prior to release of the function, in which case it requires early testing, but sometimes the testing is almost entirely focused on testing func tions that already have been released, which is in some sense a form of post- release test involvement.
Some of these strategies are more preventive, others more reactive. For example, analytical test strategies involve upfront analysis of the test basis, and tend to identify problems in the test basis prior to test execution. This allows the early -and cheap - removal of defects. That is a strength of preventive approaches.
Dynamic test strategies focus on the test execution period. Such strategies allow the location of defects and defect clusters that might have been hard to anticipate until you have the actual system in front of you. That is a strength of reactive approaches.
Rather than see the choice of strategies, particularly the preventive or reac-tive strategies, as an either/or situation, we'll let you in on the worst-kept secret of testing (and many other disciplines): There is no one best way. We suggest that you adopt whatever test approaches make the most sense in your particu-lar situation, and feel free to borrow and blend.
How do you know which strategies to pick or blend for the best chance of success? There are many factors to consider, but let us highlight a few of the most important:
• Risks: Testing is about risk management, so consider the risks and the level of risk. For a well-established application that is evolving slowly, regression is an important risk, so regression-averse strategies make sense. For a new application, a risk analysis may reveal different risks if you pick a risk-based analytical strategy.
• Skills: Strategies must not only be chosen, they must also be executed. So, you have to consider which skills your testers possess and lack. A standard- compliant strategy is a smart choice when you lack the time and skills in your team to create your own approach.
• Objectives: Testing must satisfy the needs of stakeholders to be successful. If the objective is to find as many defects as possible with a minimal amount of up-front time and effort invested - for example, at a typical independent test lab - then a dynamic strategy makes sense.
• Regulations: Sometimes you must satisfy not only stakeholders, but also reg ulators. In this case, you may need to devise a methodical test strategy that satisfies these regulators that you have met all their requirements.
• Product: Some products such as weapons systems and contract-development software tend to have well-specified requirements. This leads to synergy with a requirements-based analytical strategy.
• Business: Business considerations and business continuity are often impor tant. If you can use a legacy system as a model for a new system, you can use a model-based strategy.
We mentioned above that a good team can sometimes triumph over a situa-tion where materials, process and delaying factors are ranged against its success. However, talented execution of an unwise strategy is the equivalent of going very fast down a highway in the wrong direction. Therefore, you must make smart choices in terms of testing strategies. Furthermore, you must choose testing strategies with an eye towards the factors mentioned earlier, the schedule, budget, and feature constraints of the project and the realities of the organization and its politics.
5.3 TEST PROGRESS MONITORING AND CONTROL
1 Recall common metrics used for monitoring test preparation and execu tion. (Kl)
2 Understand and interpret test metrics for test reporting and test control (e.g., defects found and fixed and tests passed and failed). (K2)
3 Summarize the purpose and content of the test summary report docu ment according to [IEEE 829]. (K2)
In this section, we'll review techniques and metrics that are commonly used for monitoring test preparation and execution. We'll focus especially on the use and interpretation of such test metrics for reporting, controling and analyzing the test effort, including those based on defects and those based on test data. We'll also look at options for reporting test status using such metrics and other information.
As you read, remember to watch for the glossary terms defect density, failure rate, test control, test coverage, test monitoring and test report.
5.3.1 Monitoring the progress of test activities
Having developed our plans, defined our test strategies and approaches and estimated the work to be done, we must now track our testing work as we carry it out. Test monitoring can serve various purposes during the project, including the following:
• Give the test team and the test manager feedback on how the testing work is going, allowing opportunities to guide and improve the testing and the project.
• Provide the project team with visibility about the test results.
• Measure the status of the testing, test coverage and test items against the exit criteria to determine whether the test work is done.
• Gather data for use in estimating future test efforts.
Especially for small projects, the test leader or a delegated person can gather test progress monitoring information manually using documents, spreadsheets and simple databases. When working with large teams, distributed projects and long-term test efforts, we find that the efficiency and consistency of data collec-tion is aided by the use of automated tools (see Chapter 6).
One way to gather test progress information is to use the IEEE 829 test log template. While much of the information related to logging events can be use-fully captured in a document, we prefer to capture the test-by-test information in spreadsheets (see Figure 5.1).
IEEE 829 STANDARD: TEST LOG TEMPLATE
Test log identifier Activity and event entries (execution
Description (items being tested, description, procedure results,
environment in which the testing is environmental information,
conducted) anomalous events, incident report
identifiers)
In Figure 5.1, columns A and B show the test ID and the test case or test suite name. The state of the test case is shown in column C ('Warn' indicates a test that resulted in a minor failure). Column D shows the tested configuration, where the codes A, B and C correspond to test environments described in detail in the test plan. Columns E and F show the defect (or bug) ID number (from the defect-tracking database) and the risk priority number of the defect (ranging from 1, the worst, to 25, the least risky). Column G shows the initials of the tester who ran the test. Columns H through L capture data for each test related to dates, effort and duration (in hours). We have metrics for planned and actual effort and dates completed which would allow us to summarize progress against the planned schedule and budget. This spreadsheet can also be summarized in terms of the percentage of tests which have been run and the percentage of tests which have passed and failed.
Figure 5.1 might show a snapshot of test progress during the test execution period, or perhaps even at test closure if it were deemed acceptable to skip some of the tests. During the analysis, design and implementation of the tests, such a worksheet would show the state of the tests in terms of their state of development.
In addition to test case status, it is also common to monitor test progress during the test execution period by looking at the number of defects found and fixed. Figure 5.2 shows a graph that plots the total number of defects opened and closed over the course of the test execution so far.
It also shows the planned test period end date and the planned number of defects that will be found. Ideally, as the project approaches the planned end date, the total number of defects opened will settle in at the predicted number and the total number of defects closed will converge with the total number opened. These two outcomes tell us that we have found enough defects to feel comfortable that we're done testing, that we have no reason to think many more defects are lurking in the product, and that all known defects have been resolved.
Charts such as Figure 5.2 can also be used to show failure rates or defect density. When reliability is a key concern, we might be more concerned with the frequency with which failures are observed than with how many defects are causing the failures. In organizations that are looking to produce ultra-reliable
software, they may plot the number of unresolved defects normalized by the size of the product, either in thousands of source lines of code (KSLOC), func-tion points (FP) or some other metric of code size. Once the number of unre-solved defects falls below some predefined threshold - for example, three per million lines of code - then the product may be deemed to have met the defect density exit criteria.
Measuring test progress based on defects found and fixed is common and useful - if used with care. Avoid using defect metrics alone, as it is possible to achieve a flat defect find rate and to fix all the known defects by stopping any further testing, by deliberately impeding the reporting of defects and by allow-ing programmers to reject, cancel, or close defect reports without any inde-pendent review.
That said, test progress monitoring techniques vary considerably depending on the preferences of the testers and stakeholders, the needs and goals of the project, regulatory requirements, time and money constraints and other factors. In addition to the kinds of information shown in the IEEE 829 Test Log Template, Figures 5.1 and Figure 5.2, other common metrics for test progress monitoring include:
• The extent of completion of test environment preparation;
• The extent of test coverage achieved, measured against requirements, risks, code, configurations or other areas of interest;
• The status of the testing (including analysis, design and implementation) compared to various test milestones;
• The economics of testing, such as the costs and benefits of continuing test execution in terms of finding the next defect or running the next test.
As a complementary monitoring technique, you might assess the subjective level of confidence the testers have in the test items. However, avoid making important decisions based on subjective assessments alone, as people's impres-sions have a way of being inaccurate and colored by bias.
5.3.2 Reporting test status
Test progress monitoring is about gathering detailed test data; reporting test status is about effectively communicating our findings to other project stake-holders. As with test progress monitoring, in practice there is wide variability observed in how people report test status, with the variations driven by the pref-erences of the testers and stakeholders, the needs and goals of the project, reg-ulatory requirements, time and money constraints and limitations of the tools available for test status reporting. Often variations or summaries of the metrics used for test progress monitoring, such as Figure 5.1 and Figure 5.2, are used for test status reporting, too. Regardless of the specific metrics, charts and reports used, test status reporting is about helping project stakeholders under-stand the results of a test period, especially as it relates to key project goals and whether (or when) exit criteria were satisfied.
In addition to notifying project stakeholders about test results, test status reporting is often about enlightening and influencing them. This involves ana-lyzing the information and metrics available to support conclusions, recommen-dations, and decisions about how to guide the project forward or to take other actions. For example, we might estimate the number of defects remaining to be discovered, present the costs and benefits of delaying a release date to allow for further testing, assess the remaining product and project risks and offer an opinion on the confidence the stakeholders should have in the quality of the system under test.
You should think about test status reporting during the test planning and preparation periods, since you will often need to collect specific metrics during and at the end of a test period to generate the test status reports in an effective and efficient fashion. The specific data you'll want to gather will depend on your specific reports, but common considerations include the following:
• How will you assess the adequacy of the test objectives for a given test level and whether those objectives were achieved?
• How will you assess the adequacy of the test approaches taken and whether they support the achievement of the project's testing goals?
• How will you assess the effectiveness of the testing with respect to these objectives and approaches?
For example, if you are doing risk-based testing, one main test objective is to subject the important product risks to the appropriate extent of testing. Table 5.1 shows an example of a chart that would allow you to report your test cover-age and unresolved defects against the main product risk areas you identified in your risk analysis. If you are doing requirements-based testing, you could measure coverage in terms of requirements or functional areas instead of risks.
On some projects, the test team must create a test summary report. Such a report, created either at a key milestone or at the end of a test level, describes the results of a given level or phase of testing. The IEEE 829 Standard Test
Summary Report Template provides a useful guideline for what goes into such a report. In addition to including the kind of charts and tables shown earlier, you might discuss important events (especially problematic ones) that occurred during testing, the objectives of testing and whether they were achieved, the test strategy followed and how well it worked, and the overall effectiveness of the test effort.
IEEE 829 STANDARD:
TEST SUMMARY REPORT TEMPLATE
Test summary report identifier Evaluation
Summary Summary of activities
Variances Approvals
Comprehensive assessment Summary of results
5.3.3 Test control
Projects do not always unfold as planned. In fact, any human endeavor more complicated than a family picnic is likely to vary from plan. Risks become occurrences. Stakeholder needs evolve. The world around us changes. When plans and reality diverge, we must act to bring the project back under control.
In some cases, the test findings themselves are behind the divergence; for example, suppose the quality of the test items proves unacceptably bad and delays test progress. In other cases, testing is affected by outside events; for example, testing can be delayed when the test items show up late or the test environment is unavailable. Test control is about guiding and corrective actions to try to achieve the best possible outcome for the project.
The specific corrective or guiding actions depend, of course, on what we are trying to control. Consider the following hypothetical examples:
• A portion of the software under test will be delivered late, after the planned test start date. Market conditions dictate that we cannot change the release
date. Test control might involve re-prioritizing the tests so that we start testing against what is available now.
• For cost reasons, performance testing is normally run on weekday evenings during off-hours in the production environment. Due to unanticipated high demand for your products, the company has temporarily adopted an evening shift that keeps the production environment in use 18 hours a day, five days a week. Test control might involve rescheduling the performance tests for the weekend.
While these examples show test control actions that affect testing, the project team might also have to take other actions that affect others on the project. For example, suppose that the test completion date is at risk due to a high number of defect fixes that fail confirmation testing in the test environment. In this case, test control might involve requiring the programmers making the fixes to thoroughly retest the fixes prior to checking them in to the code repository for inclusion in a test build.
5.4 CONFIGURATION MANAGEMENT
1 Summarize how configuration management supports testing. (K2)
In this brief section, we'll look at how configuration management relates to and supports testing. You will come across the glossary terms configuration management and version control.
Configuration management is a topic that often perplexes new practitioners, but, if you ever have the bad luck to work as a tester on a project where this critical activity is handled poorly, you'll never forget how important it is. Briefly put, configuration management is in part about determining clearly what the items are that make up the software or system. These items include source code, test scripts, third-party software, hardware, data and both development and test documentation. Configuration management is also about making sure that these items are managed carefully, thoroughly and attentively throughout the entire project and product life cycle.
Configuration management has a number of important implications for testing. For one thing, it allows the testers to manage their testware and test results using the same configuration management mechanisms, as if they were as valuable as the source code and documentation for the system itself - which of course they are.
For another thing, configuration management supports the build process, which is essential for delivery of a test release into the test environment. Simply sending Zip archives by e-mail will not suffice, because there are too many opportunities for such archives to become polluted with undesirable contents or to harbor left-over previous versions of items. Especially in later phases of testing, it is critical to have a solid, reliable way of delivering test items that work and are the proper version.
Last but not least, configuration management allows us to map what is being tested to the underlying files and components that make it up. This is absolutely critical. For example, when we report defects, we need to report them against something, something which is version controlled. If it's not clear what we found the defect in, the programmers will have a very tough time of finding the defect in order to fix it. For the kind of test reports discussed earlier to have any meaning, we must be able to trace the test results back to what exactly we tested.
Ideally, when testers receive an organized, version-controlled test release from a change-managed source code repository, it is accompanied by a test item trans-mittal report or release notes. [IEEE 829] provides a useful guideline for what goes into such a report. Release notes are not always so formal and do not always contain all the information shown.
While our description was brief, configuration management is a topic that is as complex as test environment management. So, advanced planning is critical to making this work. During the project planning stage - and perhaps as part of your own test plan - make sure that configuration management procedures and tools are selected. As the project proceeds, the configuration process and mechanisms must be implemented, and the key interfaces to the rest of the development process should be documented. Come test execution time, this will allow you and the rest of the project team to avoid nasty surprises like testing the wrong software, receiving uninstallable builds and reporting irreproducible defects against versions of code that don't exist anywhere but in the test environment.
IEEE 829 STANDARD: TEST ITEM TRANSMITTAL REPORT TEMPLATE
Transmittal report identifier
Transmitted items
Location
Status
Approvals
5.5 RISK AND TESTING
1 Describe a risk as a possible problem that would threaten the achieve ment of one or more stakeholders' project objectives. (K2)
2 Remember that risks are determined by likelihood (of happening) and impact (harm resulting if it does happen). (Kl)
3 Distinguish between the project and product risks. (K2)
4 Recognize typical product and project risks. (Kl)
5 Describe, using examples, how risk analysis and risk management may be used for test planning. (K2)
This section covers a topic that we believe is critical to testing: risk. Let's look closely at risks, the possible problems that might endanger the objectives of the project stakeholders. We'll discuss how to determine the level of risk using like-lihood and impact. We'll see that there are risks related to the product and risks related to the project, and look at typical risks in both categories. Finally - and most important - we'll look at various ways that risk analysis and risk manage-ment can help us plot a course for solid testing
As you read this section, make sure to attend carefully to the glossary terms product risk, project risk, risk and risk-based testing.
5.5.1 Risks and levels of risk
Risk is a word we all use loosely, but what exactly is risk? Simply put, it's the possibility of a negative or undesirable outcome. In the future, a risk has some likelihood between 0% and 100%; it is a possibility, not a certainty. In the past, however, either the risk has materialized and become an outcome or issue or it has not; the likelihood of a risk in the past is either 0% or 100%.
The likelihood of a risk becoming an outcome is one factor to consider when thinking about the level of risk associated with its possible negative conse-quences. The more likely the outcome is, the worse the risk. However, likeli-hood is not the only consideration.
For example, most people are likely to catch a cold in the course of their lives, usually more than once. The typical healthy individual suffers no serious consequences. Therefore, the overall level of risk associated with colds is low for this person. But the risk of a cold for an elderly person with breathing diffi-culties would be high. The potential consequences or impact is an important consideration affecting the level of risk, too.
Remember that in Chapter 1 we discussed how system context, and espe-cially the risk associated with failures, influences testing. Here, we'll get into more detail about the concept of risks, how they influence testing, and specific ways to manage risk.
We can classify risks into project risks (factors relating to the way the work is carried out, i.e. the test project) and product risks (factors relating to what is produced by the work, i.e. the thing we are testing). We will look at product risks first.
5.5.2 Product risks
You can think of a product risk as the possibility that the system or software might fail to satisfy some reasonable customer, user, or stakeholder expecta-tion. (Some authors refer to 'product risks' as 'quality risks' as they are risks to the quality of the product.) Unsatisfactory software might omit some key func-tion that the customers specified, the users required or the stakeholders were promised. Unsatisfactory software might be unreliable and frequently fail to behave normally. Unsatisfactory software might fail in ways that cause financial or other damage to a user or the company that user works for. Unsatisfactory software might have problems related to a particular quality characteristic, which might not be functionality, but rather security, reliability, usability, main-tainability or performance.
Risk-based testing is the idea that we can organize our testing efforts in a way that reduces the residual level of product risk when the system ships. Risk-based testing uses risk to prioritize and emphasize the appropriate tests during test execution, but it's about more than that. Risk-based testing starts early in the project, identifying risks to system quality and using that knowledge of risk to guide testing planning, specification, preparation and execution. Risk-based testing involves both mitigation - testing to provide opportunities to reduce the likelihood of defects, especially high-impact defects - and contingency - testing to identify work-arounds to make the defects that do get past us less painful. Risk-based testing also involves measuring how well we are doing at finding and removing defects in critical areas, as was shown in Table 5.1. Risk-based testing can also involve using risk analysis to identify proactive opportunities to remove or prevent defects through non-testing activities and to help us select which test activities to perform.
Mature test organizations use testing to reduce the risk associated with deliv-ering the software to an acceptable level [Beizer, 1990], [Hetzel, 1988]. In the middle of the 1990s, a number of testers, including us, started to explore various techniques for risk-based testing. In doing so, we adapted well-accepted risk management concepts to software testing. Applying and refining risk assess-ment and management techniques are discussed in [Black, 2001] and [Black, 2004]. For two alternative views, see Chapter 11 of [Pol et al., 2002] and Chapter 2 of [Craig, 2002]. The origin of the risk-based testing concept can be found in Chapter 1 of [Beizer, 1990] and Chapter 2 of [Hetzel, 1988].
Risk-based testing starts with product risk analysis. One technique for risk analysis is a close reading of the requirements specification, design specifica-tions, user documentation and other items. Another technique is brainstorming with many of the project stakeholders. Another is a sequence of one-on-one or small-group sessions with the business and technology experts in the company. Some people use all these techniques when they can. To us, a team-based approach that involves the key stakeholders and experts is preferable to a purely document-based approach, as team approaches draw on the knowledge, wisdom and insight of the entire team to determine what to test and how much.
While you could perform the risk analysis by asking, 'What should we worry about?' usually more structure is required to avoid missing things. One way to provide that structure is to look for specific risks in particular product risk cat-egories. You could consider risks in the areas of functionality, localization, usability, reliability, performance and supportability. Alternatively, you could use the quality characteristics and sub-characteristics from ISO 9126 (intro-duced in Chapter 1), as each sub-characteristic that matters is subject to risks that the system might have troubles in that area. You might have a checklist of typical or past risks that should be considered. You might also want to review the tests that failed and the bugs that you found in a previous release or a similar product. These lists and reflections serve to jog the memory, forcing you to think about risks of particular kinds, as well as helping you structure the doc-umentation of the product risks.
When we talk about specific risks, we mean a particular kind of defect or failure that might occur. For example, if you were testing the calculator utility that is bundled with Microsoft Windows, you might identify 'incorrect calcula-tion' as a specific risk within the category of functionality. However, this is too
broad. Consider incorrect addition. This is a high-impact kind of defect, as everyone who uses the calculator will see it. It is unlikely, since addition is not a complex algorithm. Contrast that with an incorrect sine calculation. This is a low-impact kind of defect, since few people use the sine function on the Windows calculator. It is more likely to have a defect, though, since sine func-tions are hard to calculate.
After identifying the risk items, you and, if applicable, the stakeholders, should review the list to assign the likelihood of problems and the impact of problems associated with each one. There are many ways to go about this assignment of likelihood and impact. You can do this with all the stakeholders at once. You can have the business people determine impact and the technical people determine likelihood, and then merge the determinations. Either way, the reason for identifying risks first and then assessing their level, is that the risks are relative to each other.
The scales used to rate likelihood and impact vary. Some people rate them high, medium and low. Some use a 1-10 scale. The problem with a 1-10 scale is that it's often difficult to tell a 2 from a 3 or a 7 from an 8, unless the differences between each rating are clearly defined. A five-point scale (very high, high, medium, low and very low) tends to work well.
Given two classifications of risk levels, likelihood and impact, we have a problem, though: We need a single, aggregate risk rating to guide our testing effort. As with rating scales, practices vary. One approach is to convert each risk classification into a number and then either add or multiply the numbers to cal-culate a risk priority number. For example, suppose a particular risk has a high likelihood and a medium impact. The risk priority number would then be 6 (2 times 3).
Armed with a risk priority number, we can now decide on the various risk-mitigation options available to us. Do we use formal training for programmers or analysts, rely on cross-training and reviews or assume they know enough? Do we perform extensive testing, cursory testing or no testing at all? Should we ensure unit testing and system testing coverage of this risk? These options and more are available to us.
As you go through this process, make sure you capture the key information in a document. We're not fond of excessive documentation but this quantity of infor-mation simply cannot be managed in your head. We recommend a lightweight table like the one shown in Table 5.2; we usually capture this in a spreadsheet.
Let's finish this section with two quick tips about product risk analysis. First, remember to consider both likelihood and impact. While it might make you feel like a hero to find lots of defects, testing is also about building confidence in key
functions. We need to test the things that probably won't break but would be catastrophic if they did.
Second, risk analyses, especially early ones, are educated guesses. Make sure that you follow up and revisit the risk analysis at key project milestones. For example, if you're following a V-model, you might perform the initial analysis during the requirements phase, then review and revise it at the end of the design and implementation phases, as well as prior to starting unit test, integration test, and system test. We also recommend revisiting the risk analysis during testing. You might find you have discovered new risks or found that some risks weren't as risky as you thought and increased your con-fidence in the risk analysis.
5.5.3 Project risks
We just discussed the use of testing to manage risks to product quality. However, testing is an activity like the rest of the project and thus it is subject to risks that endanger the project. To deal with the project risks that apply to testing, we can use the same concepts we apply to identifying, prioritizing and managing product risks.
Remembering that a risk is the possibility of a negative outcome, what project risks affect testing? There are direct risks such as the late delivery of the test items to the test team or availability issues with the test environment. There are also indirect risks such as excessive delays in repairing defects found in testing or problems with getting professional system administration support for the test environment.
Of course, these are merely four examples of project risks; many others can apply to your testing effort. To discover these risks, ask yourself and other project participants and stakeholders, 'What could go wrong on the project to delay or invalidate the test plan, the test strategy and the test estimate? What are unacceptable outcomes of testing or in testing? What are the likelihoods and impacts of each of these risks?' You can see that this process is very much like the risk analysis process for products. Checklists and examples can help you identify test project risks [Black, 2004].
For any risk, product or project, you have four typical options:
• Mitigate: Take steps in advance to reduce the likelihood (and possibly the impact) of the risk.
• Contingency: Have a plan in place to reduce the impact should the risk become an outcome.
• Transfer: Convince some other member of the team or project stakeholder to reduce the likelihood or accept the impact of the risk.
• Ignore: Do nothing about the risk, which is usually a smart option only when there's little that can be done or when the likelihood and impact are low.
There is another typical risk-management option, buying insurance, which is not usually pursued for project or product risks on software projects, though it is not unheard of.
Here are some typical risks along with some options for managing them.
• Logistics or product quality problems that block tests: These can be miti gated through careful planning, good defect triage and management, and robust test design.
• Test items that won't install in the test environment: These can be mitigated through smoke (or acceptance) testing prior to starting test phases or as part of a nightly build or continuous integration. Having a defined uninstall process is a good contingency plan.
• Excessive change to the product that invalidates test results or requires updates to test cases, expected results and environments: These can be mit igated through good change-control processes, robust test design and light weight test documentation. When severe incidents occur, transference of the risk by escalation to management is often in order.
• Insufficient or unrealistic test environments that yield misleading results: One option is to transfer the risks to management by explaining the limits on test results obtained in limited environments. Mitigation - sometimes com plete alleviation - can be achieved by outsourcing tests such as performance tests that are particularly sensitive to proper test environments.
Here are some additional risks to consider and perhaps to manage:
• Organizational issues such as shortages of people, skills or training, problems with communicating and responding to test results, bad expec tations of what testing can achieve and complexity of the project team or organization.
• Supplier issues such as problems with underlying platforms or hardware, failure to consider testing issues in the contract or failure to properly respond to the issues when they arise.
• Technical problems related to ambiguous, conflicting or unprioritized requirements, an excessively large number of requirements given other project constraints, high system complexity and quality problems with the design, the code or the tests.
There may be other risks that apply to your project and not all projects are subject to the same risks. See Chapter 2 of [Black, 2001], Chapters 6 and 7 of [Black, 2004] and Chapter 3 of [Craig, 2002] for a discussion on managing project risks during testing and in the test plan.
Finally, don't forget that test items can also have risks associated with them. For example, there is a risk that the test plan will omit tests for a functional area or that the test cases do not exercise the critical areas of the system.
5.5.4 Tying it all together for risk management
We can deal with test-related risks to the project and product by applying some straightforward, structured risk management techniques. The first step is to assess or analyze risks early in the project. Like a big ocean liner, projects, espe-cially large projects, require steering well before the iceberg is in plain sight. By using a test plan template like the IEEE 829 template shown earlier, you can remind yourself to consider and manage risks during the planning phase.
It's worth repeating here that early risk analyses are educated guesses. Some of those guesses will be wrong. Make sure that you plan to re-assess and adjust your risks at regular intervals in the project and make appropriate course corrections to the testing or the project itself.
One common problem people have when organizations first adopt risk-based testing is a tendency to be excessively alarmed by some of the risks once they are clearly articulated. Do not confuse impact with likelihood or vice versa. You should manage risks appropriately, based on likelihood and impact. Triage the risks by understanding how much of your overall effort can be spent dealing with them.
It's very important to maintain a sense of perspective, a focus on the point of the exercise. As with life, the goal of risk-based testing should not be - cannot practically be - a risk-free project. What we can accomplish with risk-based testing is the marriage of testing with best practices in risk management to achieve a project outcome that balances risks with quality, features, budget and schedule.
5.6 INCIDENT MANAGEMENT
1 Recognize the content of the [IEEE 829] incident report. (Kl)
2 Write an incident report covering the observation of a failure during testing. (K3)
Let's wind down this chapter on test management with an important subject: how we can document and manage the incidents that occur during test execu-tion. We'll look at what topics we should cover when reporting incidents and defects. At the end of this section, you'll be ready to write a thorough incident report.
Keep your eyes open for the only Syllabus term in this section, incident logging.
5.6.1 What are incident reports for and how do I write good ones?
When running a test, you might observe actual results that vary from expected results. This is not a bad thing - one of the major goals of testing is to find prob-lems. Different organizations have different names to describe such situations. Commonly, they're called incidents, bugs, defects, problems or issues.
To be precise, we sometimes draw a distinction between incidents on the one hand and defects or bugs on the other. An incident is any situation where the system exhibits questionable behavior, but often we refer to an incident as a defect only when the root cause is some problem in the item we're testing.
Other causes of incidents include misconfiguration or failure of the test envi-ronment, corrupted test data, bad tests, invalid expected results and tester
mistakes. (However, in some cases the policy is to classify as a defect any inci-dent that arises from a test design, the test environment or anything else which is under formal configuration management.) We talk about incidents to indicate the possibility that a questionable behavior is not necessarily a true defect. We log these incidents so that we have a record of what we observed and can follow up the incident and track what is done to correct it.
While it is most common to find incident logging or defect reporting processes and tools in use during formal, independent test phases, you can also log, report, track, and manage incidents found during development and reviews. In fact, this is a good idea, because it gives useful information on the extent to which early - and cheaper - defect detection and removal activities are happening.
Of course, we also need some way of reporting, tracking, and managing inci-dents that occur in the field or after deployment of the system. While many of these incidents will be user error or some other behavior not related to a defect, some percentage of defects do escape from quality assurance and testing activ-ities. The defect detection percentage, which compares field defects with test defects, is an important metric of the effectiveness of the test process.
Here is an example of a DDP formula that would apply for calculating DDP for the last level of testing prior to release to the field:
defects (testers)
DDP= ----------------------------------------------------------
defects (testers) + defects (field)
It is most common to find defects reported against the code or the system itself. However, we have also seen cases where defects are reported against requirements and design specifications, user and operator guides and tests. Often, it aids the effectiveness and efficiency of reporting, tracking and manag-ing defects when the defect-tracking tool provides an ability to vary some of the information captured depending on what the defect was reported against.
In some projects, a very large number of defects are found. Even on smaller projects where 100 or fewer defects are found, you can easily lose track of them unless you have a process for reporting, classifying, assigning and managing the defects from discovery to final resolution.
An incident report contains a description of the misbehavior that was observed and classification of that misbehavior. As with any written communi-cation, it helps to have clear goals in mind when writing. One common goal for such reports is to provide programmers, managers and others with detailed information about the behavior observed and the defect. Another is to support the analysis of trends in aggregate defect data, either for understanding more about a particular set of problems or tests or for understanding and reporting the overall level of system quality. Finally, defect reports, when analyzed over a project and even across projects, give information that can lead to development and test process improvements.
When writing an incident, it helps to have the readers in mind, too. The pro-grammers need the information in the report to find and fix the defects. Before that happens, though, managers should review and prioritize the defects so that scarce testing and developer resources are spent fixing and confirmation testing the most important defects. Since some defects may be deferred - perhaps to be fixed later or perhaps, ultimately, not to be fixed at all - we should include
work-arounds and other helpful information for help desk or technical support teams. Finally, testers often need to know what their colleagues are finding so that they can watch for similar behavior elsewhere and avoid trying to run tests that will be blocked.
A good incident report is a technical document. In addition to being clear for its goals and audience, any good report grows out of a careful approach to researching and writing the report. We have some rules of thumb that can help you write a better incident report.
First, use a careful, attentive approach to running your tests. You never know when you're going to find a problem. If you're pounding on the key-board while gossiping with office mates or daydreaming about a movie you just saw, you might not notice strange behaviors. Even if you see the incident, how much do you really know about it? What can you write in your incident report?
Intermittent or sporadic symptoms are a fact of life for some defects and it's always discouraging to have an incident report bounced back as 'irrepro-ducible'. So, it's a good idea to try to reproduce symptoms when you see them and we have found three times to be a good rule of thumb. If a defect has inter-mittent symptoms, we would still report it, but we would be sure to include as much information as possible, especially how many times we tried to reproduce it and how many times it did in fact occur.
You should also try to isolate the defect by making carefully chosen changes to the steps used to reproduce it. In isolating the defect, you help guide the pro-grammer to the problematic part of the system. You also increase your own knowledge of how the system works - and how it fails.
Some test cases focus on boundary conditions, which may make it appear that a defect is not likely to happen frequently in practice. We have found that it's a good idea to look for more generalized conditions that cause the failure to occur, rather than simply relying on the test case. This helps prevent the infa-mous incident report rejoinder, 'No real user is ever going to do that.' It also cuts down on the number of duplicate reports that get filed.
As there is often a lot of testing going on with the system during a test period, there are lots of other test results available. Comparing an observed problem against other test results and known defects found is a good way to find and document additional information that the programmer is likely to find very useful. For example, you might check for similar symptoms observed with other defects, the same symptom observed with defects that were fixed in previous versions or similar (or different) results seen in tests that cover similar parts of the system.
Many readers of incident reports, managers especially, will need to under-stand the priority and severity of the defect. So, the impact of the problem on the users, customers and other stakeholders is important. Most defect-tracking systems have a title or summary field and that field should mention the impact, too.
Choice of words definitely matters in incident reports. You should be clear and unambiguous. You should also be neutral, fact-focused and impar-tial, keeping in mind the testing-related interpersonal issues discussed in Chapter 1 and earlier in this chapter. Finally, keeping the report concise helps keep people's attention and avoids the problem of losing them in the details.
As a last rule of thumb for incident reports, we recommend that you use a review process for all reports filed. It works if you have the lead tester review reports and we have also allowed testers - at least experienced ones - to review other testers' reports. Reviews are proven quality assurance techniques and incident reports are important project deliverables.
5.6.2 What goes in an incident report?
An incident report describes some situation, behavior or event that occurred during testing that requires further investigation. In many cases, an incident report consists of one or two screens - full of information gathered by a defect-tracking tool and stored in a database.
As mentioned above, you often document narrative information such as the summary, the steps to reproduce, the isolation steps tried and the impact of the problem. These fields should mention the inputs given and outputs observed, the discrepancy or variance from expectations, the different ways you could -and couldn't - make the problem recur and the impact. Classification informa-tion that a tester would provide includes the date and time of the failure, what phase of the project the failure was found in, the test case that produced the incident, references to specifications or other documents that provide informa-tion about correct behavior, the name of the tester (and perhaps the reviewer), the test environment and any additional information about the configuration of the software, system or environment. Sometimes testers classify the scope, severity and priority of the defect, though sometimes managers or a bug triage committee handle that role.
As the incident is managed to resolution, managers might assign a level of priority to the report. The change control board or bug triage committee might document the risks, costs, opportunities and benefits associated with fixing or not fixing the defect. The programmer, when fixing the defect, can capture the root cause, the phase of introduction and the phase of removal.
After the defect has been resolved, managers, programmers or others may want to capture conclusions and recommendations. Throughout the life cycle of the incident report, from discovery to resolution, the defect-tracking system should allow each person who works on the incident report to enter status and history information.
IEEE 829 STANDARD: TEST INCIDENT REPORT TEMPLATE
Test incident report identifier
Summary
Incident description (inputs, expected
results, actual results, anomalies,
date and time, procedure step,
environment, attempts to repeat,
testers and observers) Impact
5.6.3 What happens to incident reports after you file them?
As we mentioned earlier, incident reports are managed through a life cycle from discovery to resolution. The incident report life cycle is often shown as a state transition diagram (see Figure 5.3). While your defect-tracking system may use a different life cycle, let's take this one as an example to illustrate how an incident report life cycle might work.
In the incident report life cycle shown in Figure 5.3, all incident reports move through a series of clearly identified states after being reported. Some of these state transitions occur when a member of the project team completes some assigned task related to closing an incident report. Some of these state transi-tions occur when the project team decides not to repair a defect during this project, leading to the deferral of the incident report. Some of these state tran-sitions occur when an incident report is poorly written or describes behavior which is actually correct, leading to the rejection of that report.
Let's focus on the path taken by incident reports which are ultimately fixed. After an incident is reported, a peer tester or test manager reviews the report. If successful in the review, the incident report becomes opened, so now the project team must decide whether or not to repair the defect. If the defect is to be repaired, a programmer is assigned to repair it.
Once the programmer believes the repairs are complete, the incident report returns to the tester for confirmation testing. If the confirmation test fails, the incident report is re-opened and then re-assigned. Once the tester confirms a good repair, the incident report is closed. No further work remains to be done.
In any state other than rejected, deferred or closed, further work is required on the incident prior to the end of this project. In such a state, the incident report has a clearly identified owner. The owner is responsible for transitioning the incident into an allowed subsequent state. The arrows in the diagram show these allowed transitions.
In a rejected, deferred or closed state, the incident report will not be assigned to an owner. However, certain real-world events can cause an incident report to change state even if no active work is occurring on the incident report. Examples include the recurrence of a failure associated with a closed incident report and the discovery of a more serious failure associated with a deferred incident report.
Ideally, only the owner can transition the incident report from the current state to the next state and ideally the owner can only transition the incident report to an allowed next state. Most defect-tracking systems support and enforce the life cycle and life cycle rules. Good defect-tracking systems allow you to customize the set of states, the owners, and the transitions allowed to match your actual workflows. And, while a good defect-tracking system is helpful, the actual defect workflow should be monitored and supported by project and company management.
CHAPTER REVIEW
Let's review what you have learned in this chapter.
From Section 5.1, you should now be able to explain the basic ideas of test organization. You should know why independent testing is important, but also be able to analyze the potential benefits and problems associated with independent test teams. You should recognize the types of people and skills needed in a test team and recall the tasks that a tester and a test leader will carry out. You should know the glossary terms tester, test leader and test manager.
From Section 5.2, you should now understand the fundamentals of test planning and estimation. You should know the reasons for writing test plans and be able to explain how test plans relate to projects, test levels or phases, test targets and test execution. You should know which parts of the test process require special attention in test planning. You should be able to explain the justification behind various entry and exit criteria that might relate to projects, test levels or phases and test targets. You should be able to distinguish the purpose and content of test plans from that of test design specifications, test cases and test procedures, and know the IEEE 829 outline for a test plan. You should know the factors that affect the effort involved in testing, including especially test strategies (approaches) and how they affect testing. You should be able to explain how metrics, expertise and negotiation are used for estimating. You should know the glossary terms entry criteria, exit criteria, exploratory testing, test approach, test level, test plan, test procedure and test strategy.
From Section 5.3, you should be able to explain the essentials of test progress monitoring and control. You should know the common metrics that are captured, logged and used for monitoring, as well as ways to present these metrics. You should be able to analyze, interpret and explain test metrics that can be useful for reporting test status and for making decisions about how to control test progress. You should be able to explain a typical test status report and know the IEEE 829 test summary report and test log. You should know the glossary terms defect density, failure rate, test control, test coverage, test monitoring and test report.
From Section 5.4, you should now understand the basics of configuration management that relate to testing. You should be able to summarize how good configuration management helps us do our testing work better. You should know the glossary terms configuration management and version control.
From Section 5.5, you should now be able to explain how risk and testing relate. You should know that a risk is a potential undesirable or negative outcome and that most of the risks we are interested in relate to the achievement of project objectives. You should know about likelihood and impact as factors that determine the importance of a risk. You should be able to compare and contrast risks to the product (and its quality) and risks to the project itself and know typical risks to the product and project. You should be able to describe how to use risk analysis and risk management for testing and test planning. You should know the glossary terms product risk, project risk, risk and risk-based testing.
From Section 5.6, you should now understand incident logging and be able to use incident management on your projects. You should know the content of an incident report according to the IEEE 829 standard. You should be able to write a high-quality report based on test results and manage that report through its life cycle. You should know the glossary term incident logging.
SAMPLE EXAM QUESTIONS
Question 1 Why is independent testing important?
a. Independent testing is usually cheaper than testing your own work.
b. Independent testing is more effective at finding defects.
c. Independent testers should determine the processes and methodologies used.
d. Independent testers are dispassionate about whether the project succeeds or fails.
Question 2 Which of the following is among the typical tasks of a test leader?
a. Develop system requirements, design specifications and usage models.
b. Handle all test automation duties.
c. Keep tests and test coverage hidden from programmers.
d. Gather and report test progress metrics.
Question 3 According to the ISTQB Glossary, what do we mean when we call someone a test manager?
a. A test manager manages a collection of test leaders.
b. A test manager is the leader of a test team or teams.
c. A test manager gets paid more than a test leader.
d. A test manager reports to a test leader.
Question 4 What is the primary difference between the test plan, the test design specification, and the test procedure specification?
a. The test plan describes one or more levels of testing, the test design specification identifies the associated high-level test cases and a test procedure specification describes the actions for executing a test.
b. The test plan is for managers, the test design specification is for programmers and the test procedure specification is for testers who are automating tests.
c. The test plan is the least thorough, the test procedure specification is the most thorough and the test design specification is midway between the two.
d. The test plan is finished in the first third of the project, the test design specification is finished in the middle third of the project and the test procedure specification is finished in the last third of the project.
Question 5 Which of the following factors is an influence on the test effort involved in most projects?
a. Geographical separation of tester and programmers.
b. The departure of the test manager during the project.
c. The quality of the information used to develop the tests.
d. Unexpected long-term illness by a member of the project team.
Question 6 The ISTQB Foundation Syllabus establishes a fundamental test process where test planning occurs early in the project, while test execution occurs at the end. Which of the following elements of the test plan, while specified during test planning, is assessed during test execution?
a. Test tasks
b. Environmental needs
c. Exit criteria
d. Test team training
Question 7 Consider the following exit criteria which might be found in a test plan:
I No known customer-critical defects.
II All interfaces between components tested. III 100% code coverage of all units.
IVAll specified requirements satisfied.
V System functionality matches legacy system for all business rules.
Which of the following statements is true about whether these exit criteria belong in an acceptance test plan?
a. All statements belong in an acceptance test plan.
b. Only statement I belongs in an acceptance test plan.
c. Only statements I, II, and V belong in an acceptance test plan.
d. Only statements I, IV, and V belong in an acceptance test plan.
Question 8 According to the ISTQB Glossary, what is a test level?
a. A group of test activities that are organized together.
b. One or more test design specification documents.
c. A test type.
d. An ISTQB certification.
Question 9 Which of the following metrics would be most useful to monitor during test execution?
a. Percentage of test cases written.
b. Number of test environments remaining to be configured.
c. Number of defects found and fixed.
d. Percentage of requirements for which a test has been written.
Question 10 During test execution, the test manager describes the following situation to the project team: '90% of the test cases have been run. 20% of the test cases have identified defects. 127 defects have been found. 112 defects have been fixed and have passed confirmation testing. Of the remaining 15 defects, project management has decided that they do not need to be fixed prior to release.' Which of the following is the most reasonable interpretation of this test status report?
a. The remaining 15 defects should be confirmation tested prior to release.
b. The remaining 10% of test cases should be run prior to release.
c. The system is now ready for release with no further testing or development effort.
d. The programmers should focus their attention on fixing the remaining known defects prior to release.
Question 11 In a test summary report, the project's test leader makes the following statement, 'The payment processing subsystem fails to accept payments from American Express cardholders, which is considered a must-work feature for this release.' This statement is likely to be found in which of the following sections?
a. Evaluation
b. Summary of activities
c. Variances
d. Incident description
Question 12 During an early period of test execution, a defect is located, resolved and confirmed as resolved by re-testing, but is seen again later during subsequent test execution. Which of the following is a testing-related aspect of configuration management that is most likely to have broken down?
a. Traceability
b. Confirmation testing
c. Configuration control
d. Test documentation management
Question 13 You are working as a tester on a project to develop a point-of-sales system for grocery stores and other similar retail outlets. Which of the following is a product risk for such a project?
a. The arrival of a more-reliable competing product on the market.
b. Delivery of an incomplete test release to the first cycle of system test.
c. An excessively high number of defect fixes fail during re-testing.
d. Failure to accept allowed credit cards.
Question 14 A product risk analysis meeting is held during the project planning period. Which of the following determines the level of risk?
a. Difficulty of fixing related problems in code
b. The harm that might result to the user
c. The price for which the software is sold
d. The technical staff in the meeting
Question 15 You are writing a test plan using the IEEE 829 template and are currently completing the Risks and Contingencies section. Which of the following is most likely to be listed as a project risk?
a. Unexpected illness of a key team member
b. Excessively slow transaction-processing time
c. Data corruption under network congestion
d. Failure to handle a key use case
Question 16 You and the project stakeholders develop a list of product risks and project risks during the planning stage of a project. What else should you do with those lists of risks during test planning?
a. Determine the extent of testing required for the product risks and the mitigation and contingency actions required for the project risks.
b. Obtain the resources needed to completely cover each product risk with tests and transfer responsibility for the project risks to the project manager.
c. Execute sufficient tests for the product risks, based on the likelihood and impact of each product risk and execute mitigation actions for all project risks.
d. No further risk management action is required at the test planning stage.
Question 17 According to the ISTQB Glossary, a product risk is related to which of the following?
a. Control of the test project
b. The test object
c. A single test item
d. A potential negative outcome
Question 18 In an incident report, the tester makes the following statement, At this point, I expect to receive an error message explaining the rejection of this invalid input and asking me to enter a valid input. Instead the system accepts the input, displays an hourglass for between one and five seconds and finally terminates abnormally,
giving the message, "Unexpected data type: 15. Click to continue." ' This statement is likely to be found in which of the following sections of an IEEE 829 standard incident report?
a. Summary
b. Impact
c. Item pass/fail criteria
d. Incident description
Question 19 According to the ISTQB Glossary, what do we call a document that describes any event that occurred during testing which requires further investigation?
a. A bug report
b. A defect report
c. An incident report
d. A test summary report
Question 20 A product risk analysis is performed during the planning stage of the test process. During the execution stage of the test process, the test manager directs the testers to classify each defect report by the known product risk it relates to (or to 'other'). Once a week, the test manager runs a report that shows the percentage of defects related to each known product risk and to unknown risks. What is one possible use of such a report?
a. To identify new risks to system quality.
b. To locate defect clusters in product subsystems.
c. To check risk coverage by tests.
d. To measure exploratory testing.
EXERCISE: INCIDENT REPORT
Refer back to the e-mail used as an exercise in Chapter 1. Use the incident template to transform that e-mail into a well-written defect report. Make sure you take into account and resolve some of the problems with the e-mail. Feel free to use your imagination to create additional details you might need for a good defect report.
EXERCISE SOLUTION
This is a possible defect report derived from the e-mail. Notice that we have avoided repetition and emotive language. We have made sure to include sufficient evidence and the steps to reproduce the problem.
Fault report number Project number Test stage Related reports
TP1034/1 ANB10 System Test A1204/23b
TP1034/2
Brief description The client environments are not clearly identified on screen to the user and the method of logging onto a client environment allows the test and production environments to be confused. This means that if you mis-click on a .bat file when running up the client, you may open the wrong client and not realize.
Impact assessment Priority Low to medium Severity Very high
(not stopping testing proceeding) (could impact users' business)
Detailed description The client environments are not clearly identified on screen to the user and the method of logging onto a client environment allows the test and production environments to be confused. This means that if you mis-click on a .bat file when running up the client, you may open the wrong client and not realize.
This is a problem which if resolved would make testing done within the development and system test areas more efficient and it is also important for the user community.
This problem will arise for the users in two ways:
• They carry out UAT on their own machines and so the test data as well as the live clients are available in their lists and they may mis-select the client either when carrying out testing (and accidentally test in the live system) or may mis-select the test system when trading (think they have traded when they have not or raise help desk issues if they believe it to be a system problem).
• They carry out work in different countries/markets and need to know which they are operating in. Steps taken by the tester:
• We have two test environments called Systest14 and Systesti 5 and my tests were set up in Systest 15.
• To run up the clients, we have to run a .bat file for either a 14 or 15 client.
• I had started two clients, that is two exchange participants, so I could do some trading between them.
• It appears I started the first client in environment 15 but when I started the second I accidentally moved the mouse a fraction so it ran the 14 .bat file that is next to it in the Explorer file list.
The client screens do not show the environment to which you are attached and so I tried to proceed with the tests I had planned and could not get the results I expected. After some investigations I realized that I had made a mistake in the test set-up by selecting the wrong client. What caused my mistake and could it be prevented? Whereas on the server side log-on to a test environment, the environment name must be typed in and it is shown on all the panels, the log-on for the client side is done by selecting a .bat file from a list of many similarly named files. There is neither a display nor the ability to determine the client environment in which one is working.
A related problem (TP1034/2) has been raised with regard to the default country/market, which is not shown on screen and may be mis-selected. Providing a log-on screen for the client and showing the selected client on screen would prevent both these user errors.
Although this is not a functionality problem and is caused by a user mistake, it is an easy mistake to make and could have a high impact. In real-life terms it means a real user could be connected to the production system and think he or she is connected to a test system and thus make improper or illegal trades. This has happened once on the production equities trading system, when a trader entered a load of test transactions into the production system by mistake and caused considerable problems (see live incident report A1204/23b). We need therefore to prevent it happening on this system.
Impact on testing It took 15 person-hours to investigate and understand the root cause.
Evidence attached
• Screen shots of server and client environments 14 and 15
• Screen shots of equivalent live environments
• Screen shots of the country choice problem
• Recovery time stats for this problem, the country/market default problem and the equity system problem referred to.
CHAPTER 6 Tool support for testing
ou may be wishing that you had a magic tool that would automate all of the testing for you. If so, you will be disappointed. However, there are a number of very useful tools that can bring significant benefits. In this chapter we will see that there is tool support for many different aspects of software testing. We will see that success with tools is not guaranteed, even if an appropriate tool is acquired - there are also risks in using tools. There are some special considerations mentioned in the Syllabus for certain types of tool: test execution tools, performance testing tools, static analysis tools and test management tools.
6.1 TYPES OF TEST TOOL
1 Classify different types of test tools according to the test process activi ties. (K2)
2 Recognize tools that may help developers in their testing. (Kl)
In this section, we will describe the various tool types in terms of their general functionality, rather than going into lots of detail. The reason for this is that, in general, the types of tool will be fairly stable over a longer period, even though there will be new vendors in the market, new and improved tools, and even new types of tool in the coming years.
We will not mention any commercial tools in this chapter. If we did, this book would date very quickly! Tool vendors are acquired by other vendors, change their names, and change the names of the tools quite frequently, so we will not mention the names of any tools or vendors.
6.1.1 Test tool classification
The tools are grouped by the testing activities or areas that are supported by a set of tools, for example, tools that support management activities, tools to support static testing, etc. .
There is not necessarily a one-to-one relationship between a type of tool described here and a tool offered by a commercial tool vendor or an open-
source tool. Some tools perform a very specific and limited function (sometimes called a 'point solution'), but many of the commercial tools provide support for a number of different functions (tool suites or families of tools). For example a 'test management' tool may provide support for managing testing (progress monitoring), configuration management of testware, incident management, and requirements management and traceability; another tool may provide both coverage measurement and test design support.
There are some things that people can do much better or easier than a com-puter can do. For example, when you see a friend in an unexpected place, say in an airport, you can immediately recognize their face. This is an example of pattern recognition that people are very good at, but it is not easy to write soft-ware that can recognize a face.
There are other things that computers can do much better or more quickly than people can do. For example, can you add up 20 three-digit numbers quickly? This is not easy for most people to do, so you are likely to make some mistakes even if the numbers are written down. A computer does this accurately and very quickly. As another example, if people are asked to do exactly the same task over and over, they soon get bored and then start making mistakes.
The point is that it's a good idea to use computers to do things that comput-ers are really good at and that people are not very good at. So tool support is very useful for repetitive tasks - the computer doesn't get bored and will be able to exactly repeat what was done before. Because the tool will be fast, this can make those activities much more efficient and more reliable. The tools can also do things that might overload a person, such as comparing the contents of a large data file or simulating how the system would behave.
A tool that measures some aspect of software may have unexpected side-effects on that software. For example, a tool that measures timings for non-functional (performance) testing needs to interact very closely with that software in order to measure it. A performance tool will set a start time and a stop time for a given transaction in order to measure the response time, for example. But the act of taking that measurement, i.e. storing the time at those two points, could actually make the whole transaction take slightly longer than it would do if the tool wasn't measuring the response time. Of course, the extra time is very small, but it is still there. This effect is called the 'probe effect'.
Another example of the probe effect occurs with coverage tools. In order to measure coverage, the tool must first identify all of the structural elements that might be exercised to see whether a test exercises it or not. This is called 'instru-menting the code'. The tests are then run through the instrumented code so that the tool can tell (through the instrumentation) whether or not a given branch (for example) has been exercised. But the instrumented code is not the same as the real code - it also includes the instrumentation code. In theory the code is the same, but in practice, it isn't. Because different coverage tools work in slightly different ways, you may get a slightly different coverage measure on the same program because of the probe effect. For example different tools may count branches in different ways, so the percentage of coverage would be com-pared to a different total number of branches. The response time of the instru-mented code may also be significantly worse than the code without instrumentation. (There are also non-intrusive coverage tools that observe the
blocks of memory containing the object code to get a rough measurement without instrumentation, e.g. for embedded software.)
One further example of the probe effect is when a debugging tool is used to try to find a particular defect. If the code is run with the debugger, then the bug disappears; it only re-appears when the debugger is turned off (thereby making it much more difficult to find). These are sometimes known as 'Heizenbugs' (after Heizenberg's uncertainty principle).
In the descriptions of the tools below, we will indicate the tools which are more likely to be used by developers during component testing and component integration testing. For example coverage measurement tools are most often used in component testing, but performance testing tools are more often used at system testing, system integration testing and acceptance testing.
Note that for the Foundation Certificate exam, you only need to recognize the different types of tools and what they do; you do not need a detailed under-standing of them (or know how to use them).
6.1.2 Tool support for management of testing and tests
What does 'test management' mean? It could be 'the management of tests' or it could be 'managing the testing process'. The tools in this broad category provide support for either or both of these. The management of testing applies over the whole of the software development life cycle, so a test management tool could be among the first to be used in a project. A test management tool may also manage the tests, which would begin early in the project and would then continue to be used throughout the project and also after the system had been released. In practice, test management tools are typically used by special-ist testers or test managers at system or acceptance test level.
Test management tools
The features provided by test management tools include those listed below.
Some tools will provide all of these features; others may provide one or more
of the features, however such tools would still be classified as test management
tools.
Features or characteristics of test management tools include support for:
• management of tests (e.g. keeping track of the associated data for a given set of tests, knowing which tests need to run in a common environment, number of tests planned, written, run, passed or failed);
• scheduling of tests to be executed (manually or by a test execution tool);
• management of testing activities (time spent in test design, test execution, whether we are on schedule or on budget);
• interfaces to other tools, such as:
- test execution tools (test running tools);
- incident management tools;
- requirement management tools;
- configuration management tools;
• traceability of tests, test results and defects to requirements or other sources;
• logging test results (note that the test management tool does not run tests,
but could summarize results from test execution tools that the test manage-ment tool interfaces with);
• preparing progress reports based on metrics (quantitative analysis), such as:
- tests run and tests passed;
- incidents raised, defects fixed and outstanding.
This information can be used to monitor the testing process and decide what actions to take (test control), as described in Chapter 5. The tool also gives information about the component or system being tested (the test object). Test management tools help to gather, organize and communicate information about the testing on a project.
Requirements management tools
Are requirements management tools really testing tools? Some people may say they are not, but they do provide some features that are very helpful to testing. Because tests are based on requirements, the better the quality of the require-ments, the easier it will be to write tests from them. It is also important to be able to trace tests to requirements and requirements to tests, as we saw in Chapter 2.
Some requirements management tools are able to find defects in the require-ments, for example by checking for ambiguous or forbidden words, such as 'might', 'and/or', 'as needed' or '(to be decided)'.
Features or characteristics of requirements management tools include support for:
• storing requirement statements;
• storing information about requirement attributes;
• checking consistency of requirements;
• identifying undefined, missing or 'to be defined later' requirements;
• prioritizing requirements for testing purposes;
• traceability of requirements to tests and tests to requirements, functions or features;
• traceability through levels of requirements;
• interfacing to test management tools;
• coverage of requirements by a set of tests (sometimes).
Incident management tools
This type of tool is also known as a defect-tracking tool, a defect-management tool, a bug-tracking tool or a bug-management tool. However, 'incident man-agement tool' is probably a better name for it because not all of the things tracked are actually defects or bugs; incidents may also be perceived problems, anomalies (that aren't necessarily defects) or enhancement requests. Also what is normally recorded is information about the failure (not the defect) that was generated during testing - information about the defect that caused that failure would come to light when someone (e.g. a developer) begins to investigate the failure.
Incident reports go through a number of stages from initial identification and recording of the details, through analysis, classification, assignment for
fixing, fixed, re-tested and closed, as described in Chapter 5. Incident manage-ment tools make it much easier to keep track of the incidents over time.
Features or characteristics of incident management tools include support for:
• storing information about the attributes of incidents (e.g. severity);
• storing attachments (e.g. a screen shot);
• prioritizing incidents;
• assigning actions to people (fix, confirmation test, etc.);
• status (e.g. open, rejected, duplicate, deferred, ready for confirmation test, closed);
• reporting of statistics/metrics about incidents (e.g. average time open, number of incidents with each status, total number raised, open or closed).
Incident management tool functionality may be included in commercial test management tools.
Configuration management tools
An example: A test group began testing the software, expecting to find the usual fairly high number of problems. But to their surprise, the software seemed to be much better than usual this time - very few defects were found. Before they cel-ebrated the great quality of this release, they just made an additional check to see if they had the right version and discovered that they were actually testing the version from two months earlier (which had been debugged) with the tests for that earlier version. It was nice to know that this was still OK, but they weren't actually testing what they thought they were testing or what they should have been testing.
Configuration management tools are not strictly testing tools either, but good configuration management is critical for controlled testing, as was described in Chapter 5. We need to know exactly what it is that we are sup-posed to test, such as the exact version of all of the things that belong in a system. It is possible to perform configuration management activities without the use of tools, but the tools make life a lot easier, especially in complex environments.
Testware needs to be under configuration management and the same tool may be able to be used for testware as well as for software items. Testware also has different versions and is changed over time. It is important to run the correct version of the tests as well, as our earlier example shows.
Features or characteristics of configuration management tools include support for:
• storing information about versions and builds of the software and testware;
• traceability between software and testware and different versions or variants;
• keeping track of which versions belong with which configurations (e.g. oper ating systems, libraries, browsers);
• build and release management;
• baselining (e.g. all the configuration items that make up a specific release);
• access control (checking in and out).
6.1.3 Tool support for static testing
The tools described in this section support the testing activities described in Chapter 3.
Review process support tools
The value of different types of review was discussed in Chapter 3. For a very informal review, where one person looks at another's document and gives a few comments about it, a tool such as this might just get in the way. However, when the review process is more formal, when many people are involved, or when the people involved are in different geographical locations, then tool support becomes far more beneficial.
It is possible to keep track of all the information for a review process using spreadsheets and text documents, but a review tool which is designed for the purpose is more likely to do a better job. For example, one thing that should be monitored for each review is that the reviewers have not gone over the document too quickly, i.e. that the checking rate (number of pages checked per hour) was close to that recommended for that review cycle. A review process support tool could automatically calculate the checking rate and flag exceptions. The review process support tools can normally be tailored for the particular review process or type of review being done.
Features or characteristics of review process support tools include support for:
• a common reference for the review process or processes to use in different situations;
• storing and sorting review comments;
• communicating comments to relevant people;
• coordinating online reviews;
• keeping track of comments, including defects found, and providing statisti cal information about them;
• providing traceability between comments, documents reviewed and related documents;
• a repository for rules, procedures and checklists to be used in reviews, as well as entry and exit criteria;
• monitoring the review status (passed, passed with corrections, requires re- review);
• collecting metrics and reporting on key factors.
Static analysis tools (D)
The '(D)' after this (and other types of tool) indicates that these tools are more likely to be used by developers. Static analysis by tools was discussed in Chapter 3. In this section we give a summary of what the tools do.
Static analysis tools are normally used by developers as part of the develop-ment and component testing process. The key aspect is that the code (or other artefact) is not executed or run. Of course the tool itself is executed, but the source code we are interested in is the input data to the tool.
Static analysis tools are an extension of compiler technology - in fact some compilers do offer static analysis features. It is worth checking what is available from existing compilers or development environments before looking at pur-chasing a more sophisticated static analysis tool.
Static analysis can also be carried out on things other than software code, for example static analysis of requirements or static analysis of websites (for example, to assess for proper use of accessibility tags or the following of HTML standards).
Static analysis tools for code can help the developers to understand the struc-ture of the code, and can also be used to enforce coding standards. See Section 6.2.3 for special considerations when introducing static analysis tools into an organization.
Features or characteristics of static analysis tools include support to:
• calculate metrics such as cyclomatic complexity or nesting levels (which can help to identify where more testing may be needed due to increased risk);
• enforce coding standards;
• analyze structures and dependencies;
• aid in code understanding;
• identify anomalies or defects in the code (as described in Chapter 3).
Modeling tools (D)
Modeling tools help to validate models of the system or software. For example a tool can check consistency of data objects in a database and can find inconsis-tencies and defects. These may be difficult to pick up in testing - you may have tested with one data item and not realize that in another part of the database there is conflicting information related to that item. Modeling tools can also check state models or object models.
Modeling tools are typically used by developers and can help in the design of the software.
One strong advantage of both modeling tools and static analysis tools is that they can be used before dynamic tests can be run. This enables any defects that these tools can find to be identified as early as possible, when it is easier and cheaper to fix them. There are also fewer defects left to propa-gate into later stages, so development can be speeded up and there is less rework. (Of course this is difficult to show, since those defects aren't there now!)
Note that 'model-based testing tools' are actually tools that generate test inputs or test cases from stored information about a particular model (e.g. a state diagram), so are classified as test design tools (see Section 6.1.4).
Features or characteristics of modeling tools include support for:
• identifying inconsistencies and defects within the model;
• helping to identify and prioritize areas of the model for testing;
• predicting system response and behavior under various situations, such as level of load;
• helping to understand system functions and identify test conditions using a modeling language such as UML.
6.1.4 Tool support for test specification
The tools described in this section support the testing activities described in Chapter 4.
Test design tools
Test design tools help to construct test cases, or at least test inputs (which is part of a test case). If an automated oracle is available, then the tool can also con-struct the expected result, so it can actually generate test cases (rather than just test inputs).
For example, if the requirements are kept in a requirements management or test management tool, or in a Computer Aided Software Engineering (CASE) tool used by developers, then it is possible to identify the input fields, including the range of valid values. This range information can be used to identify bound-ary values and equivalence partitions. If the valid range is stored, the tool can distinguish between values that should be accepted and those that should gen-erate an error message. If the error messages are stored, then the expected result can be checked in detail. If the expected result of the input of a valid value is known, then that expected result can also be included in the test case constructed by the test design tool.
Another type of test design tool is one that helps to select combinations of possible factors to be used in testing, to ensure that all pairs of combinations of operating system and browser are tested, for example. Some of these tools may use orthogonal arrays. See [Copeland, 2003] for a description of these combi-nation techniques.
Note that the test design tool may have only a partial oracle - that is, it may know which input values are to be accepted and rejected, but it may not know the exact error message or resulting calculation for the expected result of the test. Thus the test design tool can help us to get started with test design and will identify all of the fields, but it will not do the whole job of test design for us - there will be more verification that may need to be done.
Another type of test design tool is sometimes called a 'screen scraper', a structured template or a test frame. The tool looks at a window of the graphical user interface and identifies all of the buttons, lists and input fields, and can set up a test for each thing that it finds. This means that every button will be clicked for example and every list box will be selected. This is a good start for a thorough set of tests and it can quickly and easily identify non-working buttons. However, unless the tool has access to an oracle, it may not know what should actually happen as a result of the button click.
Yet another type of test design tool may be bundled with a coverage tool. If a coverage tool has identified which branches have been covered by a set of existing tests for example, it can also identify the path that needs to be taken in order to cover the untested branches. By identifying which of the previous deci-sion outcomes need to be True or False, the tool can calculate an input value that will cause execution to take a particular path in order to increase coverage. Here the test is being designed from the code itself. In this case the presence of an oracle is less likely, so it may only be the test inputs that are constructed by the test design tool.
Features or characteristics of test design tools include support for:
• generating test input values from:
- requirements;
- design models (state, data or object);
- code;
- graphical user interfaces;
- test conditions;
• generating expected results, if an oracle is available to the tool.
The benefit of this type of tool is that it can easily and quickly identify the tests (or test inputs) that will exercise all of elements, e.g. input fields, buttons, branches. This helps the testing to be more thorough (if that is an objective of the test!)
Then we may have the problem of having too many tests and need to find a way of identifying the most important tests to run. Cutting down an unmanage-able number of tests can be done by risk analysis (see Chapter 5). Using a com-bination technique such as orthogonal arrays can also help.
Test data preparation tools
Setting up test data can be a significant effort, especially if an extensive range or volume of data is needed for testing. Test data preparation tools help in this area. They may be used by developers, but they may also be used during system or acceptance testing. They are particularly useful for per-formance and reliability testing, where a large amount of realistic data is needed.
Test data preparation tools enable data to be selected from an existing data-base or created, generated, manipulated and edited for use in tests. The most sophisticated tools can deal with a range of files and database formats.
Features or characteristics of test data preparation tools include support to:
• extract selected data records from files or databases;
• 'massage' data records to make them anonymous or not able to be identified with real people (for data protection);
• enable records to be sorted or arranged in a different order;
• generate new records populated with pseudo-random data, or data set up according to some guidelines, e.g. an operational profile;
• construct a large number of similar records from a template, to give a large set of records for volume tests, for example.
6.1.5 Tool support for test execution and logging
Test execution tools
When people think of a 'testing tool', it is usually a test execution tool that they have in mind, a tool that can run tests. This type of tool is also referred to as a 'test running tool'. Most tools of this type offer a way to get started by capturing or recording manual tests; hence they are also known as 'capture/playback' tools, 'capture/replay' tools or 'record/playback' tools. The analogy is with recording a television program, and playing it back. However, the tests are not something
which is played back just for someone to watch the tests interact with the system, which may react slightly differently when the tests are repeated. Hence cap-tured tests are not suitable if you want to achieve long-term success with a test execution tool, as is described in Section 6.2.3.
Test execution tools use a scripting language to drive the tool. The scripting language is actually a programming language. So any tester who wishes to use a test execution tool directly will need to use programming skills to create and modify the scripts. The advantage of programmable scripting is that tests can repeat actions (in loops) for different data values (i.e. test inputs), they can take different routes depending on the outcome of a test (e.g. if a test fails, go to a different set of tests) and they can be called from other scripts giving some structure to the set of tests.
When people first encounter a test execution tool, they tend to use it to 'capture/playback', which sounds really good when you first hear about it. The theory is that while you are running your manual tests, you simply turn on the 'capture', like a video recorder for a television program. However, the theory breaks down when you try to replay the captured tests - this approach does not scale up for large numbers of tests. The main reason for this is that a captured script is very difficult to maintain because:
• It is closely tied to the flow and interface presented by the GUI.
• It may rely on the circumstances, state and context of the system at the time the script was recorded. For example, a script will capture a new order number assigned by the system when a test is recorded. When that test is played back, the system will assign a different order number and reject sub sequent requests that contain the previously captured order number.
• The test input information is 'hard-coded', i.e. it is embedded in the individ ual script for each test.
Any of these things can be overcome by modifying the scripts, but then we are no longer just recording and playing back! If it takes more time to update a captured test than it would take to run the same test again manually, the scripts tend to be abandoned and the tool becomes 'shelf-ware'.
There are better ways to use test execution tools to make them work well and actually deliver the benefits of unattended automated test running. There are at least five levels of scripting and also different comparison techniques. Data-driven scripting is an advance over captured scripts but keyword-driven scripts give significantly more benefits. [Fewster and Graham, 1999], [Buwalda et al., 2001]. [Mosley and Posey, 2002] describe 'control synchronized data-driven testing'. See also Section 6.2.3.
There are many different ways to use a test execution tool and the tools themselves are continuing to gain new useful features. For example, a test exe-cution tool can help to identify the input fields which will form test inputs and may construct a table which is the first step towards data-driven scripting.
Although they are commonly referred to as testing tools, they are actually best used for regression testing (so they could be referred to as 'regression testing tools' rather than 'testing tools'). A test execution tool most often runs tests that have already been run before. One of the most significant benefits of using this type of tool is that whenever an existing system is changed (e.g. for a defect fix or an enhancement), all of the tests that were run earlier could
potentially be run again, to make sure that the changes have not disturbed the existing system by introducing or revealing a defect.
Features or characteristics of test execution tools include support for:
• capturing (recording) test inputs while tests are executed manually;
• storing an expected result in the form of a screen or object to compare to, the next time the test is run;
• executing tests from stored scripts and optionally data files accessed by the script (if data-driven or keyword-driven scripting is used);
• dynamic comparison (while the test is running) of screens, elements, links, controls, objects and values;
• ability to initiate post-execution comparison;
• logging results of tests run (pass/fail, differences between expected and actual results);
• masking or filtering of subsets of actual and expected results, for example excluding the screen-displayed current date and time which is not of interest to a particular test;
• measuring timings for tests;
• synchronizing inputs with the application under test, e.g. wait until the appli cation is ready to accept the next input, or insert a fixed delay to represent human interaction speed;
• sending summary results to a test management tool.
Test harness/unit test fram ework tools (D)
These two types of tool are grouped together because they are variants of the type of support needed by developers when testing individual components or units of software. A test harness provides stubs and drivers, which are small programs that interact with the software under test (e.g. for testing middle-ware and embedded software). See Chapter 2 for more detail of how these are used in integration testing. Some unit test framework tools provide support for object-oriented software, others for other development paradigms. Unit test frameworks can be used in agile development to automate tests in paral-lel with development. Both types of tool enable the developer to test, identify and localize any defects. The framework or the stubs and drivers supply any information needed by the software being tested (e.g. an input that would have come from a user) and also receive any information sent by the software (e.g. a value to be displayed on a screen). Stubs may also be referred to as 'mock objects'.
Test harnesses or drivers may be developed in-house for particular systems. Advice on designing test drivers can be found in [Hoffman and Strooper, 1995].
Features or characteristics of test harnesses and unit test framework tools include support for:
• supplying inputs to the software being tested;
• receiving outputs generated by the software being tested;
• executing a set of tests within the framework or using the test harness;
• recording the pass/fail results of each test (framework tools);
• storing tests (framework tools);
• support for debugging (framework tools);
• coverage measurement at code level (framework tools).
Test comparators
Is it really a test if you put some inputs into some software, but never look to see whether the software produces the correct result? The essence of testing is to check whether the software produces the correct result, and to do that, we must compare what the software produces to what it should produce. A test comparator helps to automate aspects of that comparison.
There are two ways in which actual results of a test can be compared to the expected results for the test. Dynamic comparison is where the comparison is done dynamically, i.e. while the test is executing. The other way is post-execu-tion comparison, where the comparison is performed after the test has finished executing and the software under test is no longer running.
Test execution tools include the capability to perform dynamic comparison while the tool is executing a test. This type of comparison is good for comparing the wording of an error message that pops up on a screen with the correct wording for that error message. Dynamic comparison is useful when an actual result does not match the expected result in the middle of a test - the tool can be programmed to take some recovery action at this point or go to a different set of tests.
Post-execution comparison is usually best done by a separate tool (i.e. not the test execution tool). This is the type of tool that we mean by a test compara-tor or test comparison tool and is typically a 'stand-alone' tool. Operating systems normally have file comparison tools available which can be used for post-execution comparison and often a comparison tool will be developed in-house for comparing a particular type of file or test result.
Post-execution comparison is best for comparing a large volume of data, for example comparing the contents of an entire file with the expected contents of that file, or comparing a large set of records from a database with the expected content of those records. For example, comparing the result of a batch run (e.g. overnight processing of the day's online transactions) is probably impossible to do without tool support.
Whether a comparison is dynamic or post-execution, the test comparator needs to know what the correct result is. This may be stored as part of the test case itself or it may be computed using a test oracle. See Chapter 4 for informa-tion about test oracles.
Features or characteristics of test comparators include support for:
• dynamic comparison of transient events that occur during test execution;
• post-execution comparison of stored data, e.g. in files or databases;
• masking or filtering of subsets of actual and expected results.
Coverage measurement tools (D)
How thoroughly have you tested? Coverage tools can help answer this
question.
A coverage tool first identifies the elements or coverage items that can be counted, and where the tool can identify when a test has exercised that cover-age item. At component testing level, the coverage items could be lines of code or code statements or decision outcomes (e.g. the True or False exit from an IF statement). At component integration level, the coverage item may be a call to a function or module. Although coverage can be measured at system or accept-ance testing levels, e.g. where the coverage item may be a requirement state-ment, there aren't many (if any) commercial tools at this level; there is more tool support at component testing level or to some extent at component inte-gration level.
The process of identifying the coverage items at component test level is called 'instrumenting the code', as described in Chapter 4. A suite of tests is then run through the instrumented code, either automatically using a test exe-cution tool or manually. The coverage tool then counts the number of coverage items that have been executed by the test suite, and reports the percentage of coverage items that have been exercised, and may also identify the items that have not yet been exercised (i.e. not yet tested). Additional tests can then be run to increase coverage (the tool reports accumulated coverage of all the tests run so far).
The more sophisticated coverage tools can provide support to help iden-tify the test inputs that will exercise the paths that include as-yet unexercised coverage items (or link to a test design tool to identify the unexercised items). For example, if not all decision outcomes have been exercised, the coverage tool can identify the particular decision outcome (e.g. a False exit from an IF statement) that no test has taken so far, and may then also be able to calculate the test input required to force execution to take that decision outcome.
Features or characteristics of coverage measurement tools include support for:
• identifying coverage items (instrumenting the code);
• calculating the percentage of coverage items that were exercised by a suite of tests;'
• reporting coverage items that have not been exercised as yet;
• identifying test inputs to exercise as yet uncovered items (test design tool functionality);
• generating stubs and drivers (if part of a unit test framework).
Note that the coverage tools only measure the coverage of the items that they can identify. Just because your tests have achieved 100% statement cover-age, this does not mean that your software is 100% tested!
Security tools
There are a number of tools that protect systems from external attack, for
example firewalls, which are important for any system.
Security testing tools can be used to test security by trying to break into a system, whether or not it is protected by a security tool. The attacks may focus
on the network, the support software, the application code or the underlying database.
Features or characteristics of security testing tools include support for:
• identifying viruses;
• detecting intrusions such as denial of service attacks;
• simulating various types of external attacks;
• probing for open ports or other externally visible points of attack;
• identifying weaknesses in password files and passwords;
• security checks during operation, e.g. for checking integrity of files, and intrusion detection, e.g. checking results of test attacks.
6.1.6 Tool support for performance and monitoring
The tools described in this section support testing that can be carried out on a system when it is operational, i.e. while it is running. This can be during testing or could be after a system is released into live operation.
Dynamic analysis tools (D)
Dynamic analysis tools are 'dynamic' because they require the code to be running. They are 'analysis' rather than 'testing' tools because they analyze what is happening 'behind the scenes' while the software is running (whether being executed with test cases or being used in operation).
An analogy with a car may be useful here. If you go to look at a car to buy, you might sit in it to see if is comfortable and see what sound the doors make - this would be static analysis because the car is not being driven. If you take a test drive, then you would check that the car performs as you expect (e.g. turns right when you turn the steering wheel clockwise) - this would be a test. While the car is running, if you were to check the oil pressure or the brake fluid, this would be dynamic analy-sis - it can only be done while the engine is running, but it isn't a test case.
When your PC's response time gets slower and slower over time, but is much improved after you re-boot it, this may well be due to a 'memory leak', where the programs do not correctly release blocks of memory back to the operating system. Eventually the system will run out of memory completely and stop. Re-booting restores all of the memory that was lost, so the performance of the system is now restored to its normal state.
Features or characteristics of dynamic analysis tools include support for:
• detecting memory leaks;
• identifying pointer arithmetic errors such as null pointers;
• identifying time dependencies.
These tools would typically be used by developers in component testing and component integration testing, e.g. when testing middleware, when testing security or when looking for robustness defects.
Another form of dynamic analysis for websites is to check whether each link does actually link to something else (this type of tool may be called a 'web spider'). The tool doesn't know if you have linked to the correct page, but at least it can find dead links, which may be helpful.
Performance-testing, load-testing and stress-testing tools
Performance-testing tools are concerned with testing at system level to see whether or not the system will stand up to a high volume of usage. A 'load' test checks that the system can cope with its expected number of transactions. A 'volume' test checks that the system can cope with a large amount of data, e.g. many fields in a record, many records in a file, etc. A 'stress' test is one that goes beyond the normal expected usage of the system (to see what would happen outside its design expectations), with respect to load or volume.
In performance testing, many test inputs may be sent to the software or system where the individual results may not be checked in detail. The purpose of the test is to measure characteristics, such as response times, throughput or the mean time between failures (for reliability testing).
In order to assess performance, the tool needs to generate some kind of activity on the system, and this can be done in different ways. At a very simple level the same transaction could be repeated many times, but this is not realis-tic. There are many levels of realism that could be set, depending on the tool, such as different user profiles, different types of activity, timing delays and other parameters. Adequately replicating the end-user environments or user profiles is usually key to realistic results.
Analyzing the output of a performance-testing tool is not always straightforward and it requires time and expertise. If the performance is not up to the standard expected, then some analysis needs to be performed to see where the problem is and to know what can be done to improve the performance.
Features or characteristics of performance-testing tools include support for:
• generating a load on the system to be tested;
• measuring the timing of specific transactions as the load on the system varies;
• measuring average response times;
• producing graphs or charts of responses over time.
Monitoring tools
Monitoring tools are used to continuously keep track of the status of the system in use, in order to have the earliest warning of problems and to improve service. There are monitoring tools for servers, networks, databases, security, perform-ance, website and internet usage, and applications.
Features or characteristics of monitoring tools include support for:
• identifying problems and sending an alert message to the administrator (e.g. network administrator);
• logging real-time and historical information;
• finding optimal settings;
• monitoring the number of users on a network;
• monitoring network traffic (either in real time or covering a given length of time of operation with the analysis performed afterwards).
6.1.7 Tool support for specific application areas (Kl)
In this chapter, we have described tools according to their general functional classifications. There are also further specializations of tools within these clas-sifications. For example there are web-based performance-testing tools as well as performance-testing tools for back-office systems. There are static analysis tools for specific development platforms and programming languages, since each programming language and every platform has distinct characteristics. There are dynamic analysis tools that focus on security issues, as well as dynamic analysis tools for embedded systems.
Commercial tool sets may be bundled for specific application areas such as web-based or embedded systems.
6.1.8 Tool support using other tools
The tools described in this chapter are not the only tools that a tester can make use of. You may not normally think of a word processor or a spreadsheet as a testing tool, but they are often used to store test designs, test scripts or test data. Testers may also use SQL to set up and query databases containing test data. Tools used by developers when debugging, to help localize defects and check their fixes, are also testing tools.
Developers use debugging tools when identifying and fixing defects. The debugging tools enable them to run individual and localized tests to ensure that they have correctly identified the cause of a defect and to confirm that their change to the code will indeed fix the defect.
It is a good idea to look at any type of tool available to you for ways it could be used to help support any of the testing activities. For example, testers can use Perl scripts to help compare test results.
6.2 EFFECTIVE USE OF TOOLS: POTENTIAL BENEFITS AND RISKS
1 Summarize the potential benefits and risks of test automation and tool support for testing. (K2)
2 Recognize that test execution tools can have different scripting tech niques, including data-driven and keyword-driven. (Kl)
The reason for acquiring tools to support testing is to gain benefits, by using a software program to do certain tasks that are better done by a computer than by a person.
Advice on introducing tools into an organization can be found in web arti-cles, magazines and books such as [Dustin et al., 1999], [Siteur, 2005] and [Fewster and Graham, 1999].
6.2.1 Potential benefits of using tools
There are many benefits that can be gained by using tools to support testing, whatever the specific type of tool. Benefits include:
• reduction of repetitive work;
• greater consistency and repeatability;
• objective assessment;
• ease of access to information about tests or testing.
Repetitive work is tedious to do manually. People become bored and make mistakes when doing the same task over and over. Examples of this type of repetitive work include running regression tests, entering the same test data over and over again (both of which can be done by a test execution tool), checking against coding standards (which can be done by a static analy-sis tool) or creating a specific test database (which can be done by a test data preparation tool).
People tend to do the same task in a slightly different way even when they think they are repeating something exactly. A tool will exactly reproduce what it did before, so each time it is run the result is consistent. Examples of where this aspect is beneficial include checking to confirm the correctness of a fix to a defect (which can be done by a debugging tool or test execution tool), enter-ing test inputs (which can be done by a test execution tool) and generating tests from requirements (which can be done by a test design tool or possibly a requirements management tool).
If a person calculates a value from the software or incident reports, they may inadvertently omit something, or their own subjective prejudices may lead them to interpret that data incorrectly. Using a tool means that subjective bias is removed and the assessment is more repeatable and consistently calculated. Examples include assessing the cyclomatic complexity or nesting levels of a component (which can be done by a static analysis tool), coverage (coverage measurement tool), system behavior (monitoring tools) and incident statistics (test management tool).
Having lots of data doesn't mean that information is communicated. Information presented visually is much easier for the human mind to take in and interpret. For example, a chart or graph is a better way to show informa-tion than a long list of numbers - this is why charts and graphs in spread-sheets are so useful. Special purpose tools give these features directly for the information they process. Examples include statistics and graphs about test progress (test execution or test management tool), incident rates (incident management or test management tool) and performance (performance testing tool).
In addition to these general benefits, each type of tool has specific benefits relating to the aspect of testing that the particular tool supports. These benefits are normally prominently featured in the sales information available for the type of tool. It is worth investigating a number of different tools to get a general view of the benefits.
6.2.2 Risks of using tools
Although there are significant benefits that can be achieved using tools to support testing activities, there are many organizations that have not achieved the benefits they expected.
Simply purchasing a tool is no guarantee of achieving benefits, just as buying membership in a gym does not guarantee that you will be fitter. Each type of tool requires investment of effort and time in order to achieve the potential benefits.
There are many risks that are present when tool support for testing is intro-duced and used, whatever the specific type of tool. Risks include:
• unrealistic expectations for the tool;
• underestimating the time, cost and effort for the initial introduction of a tool;
• underestimating the time and effort needed to achieve significant and con tinuing benefits from the tool;
• underestimating the effort required to maintain the test assets generated by the tool;
• over-reliance on the tool.
Unrealistic expectations may be one of the greatest risks to success with tools. The tools are only software and we all know that there are many problems with any kind of software! It is important to have clear objectives for what the tool can do and that those objectives are realistic.
Introducing something new into an organization is seldom straightforward. Having purchased a tool, you will want to move from opening the box to having a number of people being able to use the tool in a way that will bring benefits. There will be technical problems to overcome, but there will also be resistance from other people - both need to be addressed in order to succeed in introduc-ing a tool.
Think back to the last time you did something new for the very first time (learning to drive, riding a bike, skiing). Your first attempts were unlikely to be very good but with more experience you became much better. Using a testing tool for the first time will not be your best use of the tool either. It takes time to develop ways of using the tool in order to achieve what is possible. Fortunately there are some short-cuts (e.g. reading books and articles about other people's experiences and learning from them). See also Section 6.3 for more detail on introducing a tool into an organization.
Insufficient planning for maintenance of the assets that the tool pro-duces is a strong contributor to tools that end up as 'shelf-ware', along with the previously listed risks. Although particularly relevant for test exe-cution tools, planning for maintenance is also a factor with other types of tool.
Tools are definitely not magic! They can do very well what they have been designed to do (at least a good quality tool can), but they cannot do everything. A tool can certainly help, but it does not replace the intelligence needed to know how best to use it, and how to evaluate current and future uses of the tool. For example, a test execution tool does not replace the need for good test design and should not be used for every test - some tests are still better
executed manually. A test that takes a very long time to automate and will not be run very often is better done manually.
This list of risks is not exhaustive. Two other important factors are:
• the skill needed to create good tests;
• the skill needed to use the tools well, depending on the type of tool.
The skills of a tester are not the same as the skills of the tool user. The tester concentrates on what should be tested, what the test cases should be and how to prioritize the testing. The tool user concentrates on how best to get the tool to do its job effectively and how to give increasing benefit from tool use.
6.2.3 Special considerations for some types of tools
Test execution tools
A 'capture/playback tool' is a misleading term, although it is in common use. Capture/playback is one mode of using a test execution tool and probably the worst way to use it!
In order to know what tests to execute and how to run them, the test execu-tion tool must have some way of knowing what to do - this is the script for the tool. But since the tool is only software, the script must be completely exact and unambiguous to the computer (which has no common sense). This means that the script becomes a program, written in a programming language. The script-ing language may be specific to a particular tool, or it may be a more general language. Scripting languages are not used just by test execution tools, but the scripts used by the tool are stored electronically to be run when the tests are executed under the tool's control.
There are tools that can generate scripts by identifying what is on the screen rather than by capturing a manual test, but they still generate scripts to be used in execution; they are not script-free.
There are different levels of scripting. Five are described in [Fewster and Graham, 1999]:
• linear scripts (which could be created manually or captured by recording a manual test);
• structured scripts (using selection and iteration programming structures);
• shared scripts (where a script can be called by other scripts so can be re-used - shared scripts also require a formal script library under configuration man agement);
• data-driven scripts (where test data is in a file or spreadsheet to be read by a control script);
• keyword-driven scripts (where all of the information about the test is stored in a file or spreadsheet, with a number of control scripts that implement the tests described in the file).
Capturing a manual test seems like a good idea to start with, particularly if you are currently running tests manually anyway. But a captured test (a linear script) is not a good solution, for a number of reasons, including:
• The script doesn't know what the expected result is until you program it in - it only stores inputs that have been recorded, not test cases.
• A small change to the software may invalidate dozens or hundreds of scripts.
• The recorded script can only cope with exactly the same conditions as when it was recorded. Unexpected events (e.g. a file that already exists) will not be interpreted correctly by the tool.
However, there are some times when capturing test inputs (i.e. record-ing a manual test) is useful. For example, if you are doing exploratory testing or if you are running unscripted tests with experienced business users, it can be very helpful simply to log everything that is done, as an audit trail. This serves as a form of documentation of what was tested (although analyzing it may not be easy). This audit trail can also be very useful if a failure occurs which cannot be easily reproduced - the recording of the specific failure can be played to the developer to see exactly what sequence caused the problem.
Captured test inputs can be useful in the short term, where the context remains valid. Just don't expect to replay them as regression tests (when the context of the test may be different). Captured tests may be acceptable for a few dozen tests, where the effort to update them when the software changes is not very large. Don't expect a linear-scripting approach to scale to hundreds or thousands of tests.
So capturing tests does have a place, but it is not a large place in terms of automating test execution.
Data-driven scripts allow the data, i.e. the test inputs and expected out-comes, to be stored separately from the script. This has the advantage that a tester who doesn't know how to use a scripting language can populate a file or spreadsheet with the data for a specific test. This is particularly useful when there are a large number of data values that need to be tested using the same control script.
Keyword-driven scripts include not just data but also keywords in the data file or spreadsheet. This enables a tester (who is not a script programmer) to devise a great variety of tests (not just the test input data for essentially the same test, as in data-driven scripts). The tester needs to know what keywords are currently available to use (by someone having written a script for it) and what data the keyword is expecting, but the tester can then write tests, not just test data. The tester can also request additional keywords to be added to the available programmed set of scripts as needed. Keywords can deal with both test inputs and expected outcomes.
Of course, someone still needs to be able to use the tool directly and be able to program in the tool's scripting language in order to write and debug the scripts that will use the data tables or keyword tables. A small number of automation specialists can support a larger number of testers, who then don't need to learn to be script programmers (unless they want to).
The data files (data-driven or keyword-driven) include the expected results for the tests. The actual results from each test run also need to be stored, at least until they are compared to the expected results and any differences are logged.
More information on data-driven and keyword-driven scripting can be found in [Fewster and Graham, 1999].
Performance testing tools
Performance testing is developing into a specialized discipline of its own. With functional testing, the types of defect that we are looking for are functionality -does the system or component produce the correct result for given inputs? In performance testing, we are not normally concerned so much with functional correctness, but with non-functional quality characteristics. When using a per-formance testing tool we are looking at the transaction throughput, the degree of accuracy of a given computation, the computer resources being used for a given level of transactions, the time taken for certain transactions or the number of users that can use the system at once.
In order to get the best from a performance-testing tool, it is important to understand what the tool can and cannot do for you. Although this is true for other types of tool as well, there are particular issues with performance-testing tools, including:
• the design of the load to be generated by the tool (e.g. random input or according to user profiles);
• timing aspects (e.g. inserting delays to make simulated user input more realistic);
• the length of the test and what to do if a test stops prematurely;
• narrowing down the location of a bottleneck;
• exactly what aspects to measure (e.g. user interaction level or server level);
• how to present the information gathered.
Static analysis tools
Static analysis tools are very useful to developers, as they can identify potential problems in code before the code is executed and they can also help to check that the code is written to coding standards.
When a static analysis tool is first introduced, there can be some problems. For example, if the tool checks the current coding standard against code written several years ago, there may be a number of things found in the old code that fail to meet the new coding standard now in place. If the old code has been working well for years, it is probably not a good idea to change it just to satisfy the new coding standard (unless changes are necessary for some other reason). There is a risk that the changes to meet the new standard may have inadvertent side-effects which may not be picked up by regression testing.
Static analysis tools can generate a large number of messages, for example by finding the same thing every few lines. This can be rather annoying, espe-cially if the things found are not considered important now, for example warn-ings rather than potential defects.
The aim of the static analysis tool is to produce code that will be easier to maintain in the future, so it would be a good idea to implement higher stan-dards on new code that is still being tested, before it is released into use, but to allow older code to be less stringently checked. There is still a risk that the changes to conform to the new standard will introduce an unexpected side-effect, but there is a much greater likelihood that it will be found in testing and there is time to fix it before the system is released.
A filter on the output of. the static analysis tool could eliminate some of the less important messages and make the more important messages more likely to be noticed and fixed.
Test management tools
Test management tools can provide a lot of useful information, but the informa-tion as produced by the tool may not be in the form that will be most effective within your own context. Some additional work may be needed to produce interfaces to other tools or a spreadsheet in order to ensure that the informa-tion is communicated in the most effective way.
A report produced by a test management tool (either directly or indirectly through another tool or spreadsheet) may be a very useful report at the moment, but may not be useful in three or six months. It is important to monitor the information produced to ensure it is the most relevant now.
It is important to have a defined test process before test management tools are introduced. If the testing process is working well manually, then a test management tool can help to support the process and make it more efficient. If you adopt a test management tool when your own testing processes are immature, one option is to follow the standards and processes that are assumed by the way the tool works. This can be helpful; but it is not necessary to follow the vendor-specific processes. The best approach is to define your own processes, taking into account the tool you will be using, and then adapt the tool to provide the greatest benefit to your organization.
6.3 INTRODUCING A TOOL INTO AN ORGANIZATION
1 State the main principles of introducing a tool into an organization. (Kl)
2 State the goals of a proof-of-concept or piloting phase for tool evalua tion. (Kl)
3 Recognize that factors other than simply acquiring a tool are required for good tool support. (Kl)
6.3.1 Main principles
The place to start when introducing a tool into an organization is not with the tool - it is with the organization. In order for a tool to provide benefit, it must match a need within the organization, and solve that need in a way that is both effective and efficient. The tool should help to build on the strengths of the organization and address its weaknesses. The organization needs to be ready for the changes that will come with the new tool. If the current testing practices are not good and the organization is not mature, then it is generally more cost-effective to improve testing practices rather than to try to find tools to support poor practices. Automating chaos just gives faster chaos!
Of course, we can sometimes improve our own processes in parallel with introducing a tool to support those practices and we can pick up some good ideas for improvement from the ways that the tools work. However, be aware that the tool should not take the lead, but should provide support to what your organization defines.
The following factors are important in selecting a tool:
• assessment of the organization's maturity (e.g. readiness for change);
• identification of the areas within the organization where tool support will help to improve testing processes;
• evaluation of tools against clear requirements and objective criteria;
• proof-of-concept to see whether the product works as desired and meets the requirements and objectives defined for it;
• evaluation of the vendor (training, support and other commercial aspects) or open-source network of support;
• identifying and planning internal implementation (including coaching and mentoring for those new to the use of the tool).
6.3.2 Pilot project
One of the ways to do a proof-of-concept is to have a pilot project as the first thing done with a new tool. This will use the tool in earnest but on a small scale, with sufficient time to explore different ways of using the tool. Objectives should be set for the pilot in order to assess whether or not the concept is proven, i.e. that the tool can accomplish what is needed within the current orga-nizational context.
A pilot tool project should expect to encounter problems - they should be solved in ways that can be used by everyone later on. The pilot project should experiment with different ways of using the tool. For example, different settings for a static analysis tool, different reports from a test management tool, differ-ent scripting and comparison techniques for a test execution tool or different load profiles for a performance-testing tool.
The objectives for a pilot project for a new tool are:
• to learn more about the tool (more detail, more depth);
• to see how the tool would fit with existing processes or documentation, how those would need to change to work well with the tool and how to use the tool to streamline existing processes;
• to decide on standard ways of using the tool that will work for all potential users (e.g. naming conventions, creation of libraries, defining modularity, where different elements will be stored, how they and the tool itself will be maintained);
• to evaluate the pilot project against its objectives (have the benefits been achieved at reasonable cost?).
6.3.3 Success factors
Success is not guaranteed or automatic when implementing a testing tool, but many organizations have succeeded. Here are some of the factors that have contributed to success:
• incremental roll-out (after the pilot) to the rest of the organization;
• adapting and improving processes, testware and tool artefacts to get the best fit and balance between them and the use of the tool;
• providing adequate training, coaching and mentoring of new users;
• defining and communicating guidelines for the use of the tool, based on what was learned in the pilot;
• implementing a continuous improvement mechanism as tool use spreads through more of the organization;
• monitoring the use of the tool and the benefits achieved and adapting the use of the tool to take account of what is learned.
More information and advice about selecting and implementing tools can be found in [Fewster and Graham, 1999] and [Dustin et al., 1999].
CHAPTER REVIEW
Let's review what you have learned in this chapter.
From Section 6.1, you should now be able to classify different types of test tools according to the test process activities that they support. You should also recognize the tools that may help developers in their testing (shown by '(D)' below). In addition to the list below, you should recognize that there are tools that support specific application areas and that general-purpose tools can also be used to support testing. The tools you should now recognize are:
Tools that support the management of testing and tests:
- test management tool;
- requirements management tool;
- incident management tool;
- configuration management tool.
Tools that support static testing:
- review process support tool;
- static analysis tool (D);
- modeling tool (D).
Tools that support test specification:
- test design tool;
- test data preparation tool.
Tools that support test execution and logging:
- test execution tool;
- test harness and unit test framework tool (D);
- test comparator;
- coverage measurement tool (D);
- security tool.
Tools that support performance and monitoring:
- dynamic analysis tool;
- performance-testing, load-testing and stress-testing tool;
- monitoring tool.
In addition to the tools already listed, you should know the glossary terms debugging tool, driver, probe effect and stub.
From Section 6.2, you should be able to summarize the potential benefits and potential risks of tool support for testing in general. You should recognize that some tools have special considerations, including test execution tools, per-formance-testing tools, static analysis tools and test management tools. You should know the glossary terms data-driven testing, keyword-driven testing and scripting language and recognize these as associated with test execution tools.
From Section 6.3, you should be able to state the main principles of introduc-ing a tool into an organization (e.g. assessing organizational maturity, clear requirements and objective criteria, proof-of-concept, vendor evaluation, coaching and mentoring). You should be able to state the goals of a proof-of-concept or piloting phase for tool evaluation (e.g. learn about the tool, assess fit with current practices, decide on standards, assess benefits). You should recog-nize that simply acquiring a tool is not the only factor in achieving good tool
support; there are many other factors that are important for success (e.g. incre-mental roll-out, adapting processes, training and coaching, defining usage guidelines, learning lessons and monitoring benefits). There are no specific def-initions for this section.
SAMPLE EXAM QUESTIONS
Question 1 Which tools help to support static testing?
a. Static analysis tools and test execution tools.
b. Review process support tools, static analysis tools and coverage measurement tools.
c. Dynamic analysis tools and modeling tools.
d. Review process support tools, static analysis tools and modeling tools.
Question 2 Which test activities are supported by test harness or unit test framework tools?
a. Test management and control.
b. Test specification and design.
c. Test execution and logging.
d. Performance and monitoring.
Question 3 What are the potential benefits from using tools in general to support testing?
a. Greater quality of code, reduction in the number of testers needed, better objectives for testing.
b. Greater repeatability of tests, reduction in repetitive work, objective assessment.
c. Greater responsiveness of users, reduction of tests run, objectives not necessary.
d. Greater quality of code, reduction in paperwork, fewer objections to the tests.
Question 4 What is a potential risk in using tools to support testing?
a. Unrealistic expectations, expecting the tool to do too much.
b. Insufficient reliance on the tool, i.e. still doing manual testing when a test execution tool has been purchased.
c. The tool may find defects that aren't there.
d. The tool will repeat exactly the same thing it did the previous time.
Question 5 Which of the following are advanced scripting techniques for test execution tools?
a. Data-driven and keyword-driven
b. Data-driven and capture-driven
c. Capture-driven and keyhole-driven
d. Playback-driven and keyword-driven
Question 6 Which of the following would NOT be done as part of selecting a tool for an organization?
a. Assess organizational maturity, strengths and weaknesses.
b. Roll out the tool to as many users as possible within the organization.
c. Evaluate the tool features against clear requirements and objective criteria.
d. Identify internal requirements for coaching and mentoring in the use of the tool.
Question 7 Which of the following is a goal for a proof-of-concept or pilot phase for tool evaluation?
a. Decide which tool to acquire.
b. Decide on the main objectives and requirements for this type of tool.
c. Evaluate the tool vendor including training, support and commercial aspects.
d. Decide on standard ways of using, managing, storing and maintaining the tool and the test
assets.
MOCK EXAM
On the real exam, you will have 60 minutes to work through 40 questions of approximately the same difficulty mix and Syllabus distribution as shown in the following mock exam. After you have taken this mock exam, check your answers with the answer key.
Question 1 What is a key characteristic of specification-based testing techniques?
a. Tests are derived from information about how the software is constructed.
b. Tests are derived from models (formal or informal) that specify the problem to be solved by the software or its components.
c. Tests are derived based on the skills and experience of the tester.
d. Tests are derived from the extent of the coverage of structural elements of the system or components.
Question 2 An exhaustive test suite would include:
a. All combinations of input values and preconditions.
b. All combinations of input values and output values.
c. All pairs of input value and preconditions.
d. All states and state transitions.
Question 3 Which statement about testing is true?
a. Testing is started as early as possible in the life cycle.
b. Testing is started after the code is written so that we have a system with which to work.
c. Testing is most economically done at the end of the life cycle.
d. Testing can only be done by an independent test team.
Question 4 For a test procedure that is checking modifications of customers on a database, which two steps below would be the lowest priority if we didn't have time to execute all of the steps?
1 Open database and confirm existing customer
2 Change customer's marital status from single to married
3 Change customer's street name from Parks Road to Park Road
4 Change customer's credit limit from 500 to 750
5 Replace customer's first name with exactly the same first name
6 Close the customer record and close the database
a. Tests 1 and 4
b. Tests 2 and 3
c. Tests 5 and 6
d. Tests 3 and 5
Question 5 Consider the following list of either product or project risks:
I An incorrect calculation of fees might shortchange the organization.
II A vendor might fail to deliver a system component on time.
IIIA defect might allow hackers to gain administrative privileges.
IVA skills gap might occur in a new technology used in the system.
V A defect-prioritization process might overload the development team.
Which of the following statements is true?
a. I is primarily a product risk and II, III, IV and V are primarily project risks.
b. II and V are primarily product risks and I, III and V are primarily project risks.
c. I and III are primarily product risks, while II, IV and V are primarily project risks.
d. Ill and V are primarily product risks, while I, II and IV are primarily project risks.
Question 6 Consider the following statements about regression tests:
I They may usefully be automated if they are well designed.
II They are the same as confirmation tests (re-tests).
III They are a way to reduce the risk of a change having an adverse affect elsewhere in the system.
IVThey are only effective if automated. Which pair of statements is true?
a. I and II
b. I and III
c. II and III
d. II and IV
Question 7 Which of the following could be used to assess the coverage achieved for structure-based (white-box) test techniques?
V Decision outcomes exercised W Partitions exercised
X Boundaries exercised
Y Conditions or multiple conditions exercised Z Statements exercised
a. V,WorY
b. WXorY
c. V,YorZ
d. W,XorZ
Question 8 Review the following portion of an incident report.
1 I place any item in the shopping cart.
2 I place any other (different) item in the shopping cart.
3 I remove the first item from the shopping cart, but leave the second item in the cart.
4 I click the < Checkout > button.
5 I expect the system to display the first checkout screen. Instead, it gives the pop-up error message, 'No items in shopping cart. Click <Okay> to continue shopping.'
6 I click < Okay >.
7 I expect the system to return to the main window to allow me to continue adding and removing items from the cart. Instead, the browser terminates.
8 The failure described in steps 5 and 7 occurred in each of three attempts to perform steps 1,2,3,4 and 6.
Assume that no other narrative information is included in the report. Which of the following important aspects of a good incident report is missing from this incident report?
a. The steps to reproduce the failure.
b. The summary.
c. The check for intermittence.
d. The use of an objective tone.
Question 9 Which of the following are benefits and which are risks of using tools to support testing?
1 Over-reliance on the tool
2 Greater consistency and repeatability
3 Objective assessment
4 Unrealistic expectations
5 Underestimating the effort required to maintain the test assets generated by the tool
6 Ease of access to information about tests or testing
7 Repetitive work is reduced
a. Benefits: 3,4,6 and 7. Risks: 1,2 and 5
b. Benefits: 1,2,3 and 7, Risks: 4,5 and 6
c. Benefits: 2,3,6 and 7. Risks: 1,4 and 5
d. Benefits: 2,3,5 and 6. Risks: 1,4 and 7
Question 10 Which of the following encourages objective testing?
a. Unit testing
b. System testing
c. Independent testing
d. Destructive testing
Question 11 Of the following statements about reviews of specifications, which statement is true?
a. Reviews are not generally cost effective as the meetings are time consuming and require preparation and follow up.
b. There is no need to prepare for or follow up on reviews.
c. Reviews must be controlled by the author.
d. Reviews are a cost effective early static test on the system.
Question 12 Consider the following list of test process activities:
I Analysis and design
II Test closure activities
IIIEvaluating exit criteria and reporting IVPlanning and control
V Implementation and execution
Which of the following places these in their logical sequence?
a. I, II, III, IV and V
b. IV, I, V, III and II.
c. IV, I, V,II and III.
d. I, IV, V HI and II.
Question 13 Test objectives vary between projects and so must be stated in the test plan. Which one of the following test objectives might conflict with the proper tester mindset?
a. Show that the system works before we ship it.
b. Find as many defects as possible.
c. Reduce the overall level of product risk.
d. Prevent defects through early involvement.
Question 14 Which test activities are supported by test data preparation tools?
a. Test management and control
b. Test specification and design
c. Test execution and logging
d. Performance and monitoring
Question 15 If you are flying with an economy ticket, there is a possibility that you may get upgraded to business class, especially if you hold a gold card in the airline's frequent flyer program. If you don't hold a gold card, there is a possibility that you will get 'bumped' off the flight if it is full and you check in late. This is shown in Figure 7.1. Note that each box (i.e. statement) has been numbered.
Three tests have already been run:
Test 1: Gold card holder who gets upgraded to
business class
Test 2: Non-gold card holder who stays in economy
Test 3: A person who is bumped from the flight
What additional tests would be needed to achieve 100% decision coverage?
a. A gold card holder who stays in economy and a non-gold card holder who gets upgraded to business class.
b. A gold card holder and a non-gold card holder who are both upgraded to business class.
c. A gold card holder and a non-gold card holder who both stay in economy class.
d. A gold card holder who is upgraded to business class and a non-gold card holder who stays in economy class.
Question 16 Consider the following types of tools:
V Test management tools W Static analysis tools
X Modeling tools
Y Dynamic analysis tools
Z Performance testing tools
Which of the following of these tools is most likely to be used by developers?
a. W,XandY
b. VYandZ
c. V,WandZ
d. X,YandZ
Question 17 What is a test condition?
a. An input, expected outcome, precondition and postcondition
b. The steps to be taken to get the system to a given point
c. Something that can be tested
d. A specific state of the software, e.g. before a test can be run
Question 18 Which of the following is the most important difference between the metrics-based approach and the expert-based approach to test estimation?
a. The metrics-based approach is more accurate than the expert-based approach.
b. The metrics-based approach uses calculations from historical data while the expert-based approach relies on team wisdom.
c. The metrics-based approach can be used to verify an estimate created using the expert-based approach, but not vice versa.
d. The expert-based approach takes longer than the metrics-based approach.
Question 19 If the temperature falls below 18 degrees, the heating is switched on. When the temperature reaches 21 degrees, the heating is switched off. What is the minimum set of test input values to cover all valid equivalence partitions?
a. 15,19 and 25 degrees
b. 17,18,20 and 21 degrees
c. 18,20 and 22 degrees
d. 16 and 26 degrees
Question 20 Which of these statements about functional testing is true?
a. Structural testing is more important than functional testing as it addresses the code.
b. Functional testing is useful throughout the life cycle and can be applied by business analysts, testers, developers and users.
c. Functional testing is more powerful than static testing as you actually run the system and see what happens.
d. Inspection is a form of functional testing.
Question 21 What is the purpose of confirmation testing?
a. To confirm the users' confidence that the system will meet their business needs.
b. To confirm that a defect has been fixed correctly.
c. To confirm that no unexpected changes have been introduced or uncovered as a result of changes made.
d. To confirm that the detailed logic of a component conforms to its specification.
Question 22 Which success factors are required for good tool support within an organization?
a. Acquiring the best tool and ensuring that all testers use it.
b. Adapting processes to fit with the use of the tool and monitoring tool use and benefits.
c. Setting ambitious objectives for tool benefits and aggressive deadlines for achieving them.
d. Adopting practices from other successful organizations and ensuring that initial ways of using the tool are maintained.
Question 23 Which of the following best describes integration testing?
a. Testing performed to expose faults in the interfaces and in the interaction between integrated components.
b. Testing to verify that a component is ready for integration.
c. Testing to verify that the test environment can be integrated with the product.
d. Integration of automated software test suites with the product.
Question 24 According to the ISTQB Glossary, debugging:
a. Is part of the fundamental testing process.
b. Includes the repair of the cause of a failure.
c. Involves intentionally adding known defects.
d. Follows the steps of a test procedure.
Question 25 Which of the following could be a root cause of a defect in financial software in which an incorrect interest rate is calculated?
a. Insufficient funds were available to pay the interest rate calculated.
b. Insufficient calculations of compound interest were included.
c. Insufficient training was given to the developers concerning compound interest calculation rules.
d. Inaccurate calculators were used to calculate the expected results.
Question 26 Assume postal rates for 'light letters' are:
$0.25 up to 10 grams; $0.35 up to 50 grams; $0.45 up to 75 grams; $0.55 up to 100 grams.
Which test inputs (in grams) would be selected using boundary value analysis?
a. 0,9,19,49,50,74,75, 99,100
b. 10,50,75,100,250,1000
c. 0,1,10,11,50,51,75,76,100,101
d. 25,26,35,36,45,46,55,56
Question 27 Consider the following decision table.
Given this decision table, what is the expected result for the following test cases?
TCI: A 26-year-old on business but with violations or accidents on his driving record
TC2: A 62-year-old tourist with a clean driving record
a. TCI: Don't supply car; TC2: Supply car with premium charge.
b. TCI: Supply car with premium charge; TC2: Supply car with no premium charge.
c. TCI: Don't supply car; TC2: Supply car with no premium charge.
d. TCI: Supply car with premium charge; TC2: Don't supply car.
Question 28 What is exploratory testing?
a. The process of anticipating or guessing where defects might occur.
b. A systematic approach to identifying specific equivalent classes of input.
c. The testing carried out by a chartered engineer.
d. Concurrent test design, test execution, test logging and learning.
Question 29 What does it mean if a set of tests has achieved 90% statement coverage?
a. 9 out of 10 decision outcomes have been exercised by this set of tests.
b. 9 out of 10 statements have been exercised by this set of tests.
c. 9 out of 10 tests have been run on this set of software.
d. 9 out of 10 requirements statements about the software are correct.
Question 30 A test plan is written specifically to describe a level of testing where the primary goal is establishing confidence in the system. Which of the following is a likely name for this document?
a. Master test plan
b. System test plan
c. Acceptance test plan
d. Project plan
Question 31 Requirement 24.3. A 'Postage Assistant' will calculate the amount of postage due for letters and small packages up to 1 kilogram in weight. The inputs are: the type of item (letter, book or other package) and the weight in grams. Which of the following conform to the required contents of a test case?
a. Test the three types of item to post and three different weights [Req 24.3]
b. Test 1: letter, 10 grams, postage €0.25. Test 2: book, 500 grams, postage €1.00. Test 3: package, 999 gram, postage €2.53 [Req 24.3]
c. Test 1: letter, 10 grams to Belgium. Test 2: book 500 grams to USA. Test 3: package, 999 grams to South Africa [Req 24.3]
d. Test 1: letter 10 grams, Belgium, postage €0.25. Test 2: package 999 grams to South Africa, postage €2.53
Question 32 What is the best description of static analysis?
a. The analysis of batch programs
b. The reviewing of test plans
c. The analysis of program code or other software artifacts
d. The use of black-box testing
Question 33 System test execution on a project is planned for eight weeks. After a week of testing, a tester suggests that the test objective stated in the test plan of 'finding as many defects as possible during system test' might be more closely met by redirecting the test effort according to which test principle?
a. Impossibility of exhaustive testing.
b. Importance of early testing.
c. The absence of errors fallacy.
d. Defect clustering.
Question 34 Consider the following activities that might relate to configuration management:
I Identify and document the characteristics of a test item
II Control changes to the characteristics of a test item
III Check a test item for defects introduced by a change
IVRecord and report the status of changes to test items
V Confirm that changes to a test item fixed a defect Which of the following statements is true?
a. Only I is a configuration management task.
b. All are configuration management tasks.
c. I, II and III are configuration management tasks.
d. I, II and IV are configuration management tasks.
Question 35 Consider the following state transition diagram.
Given this diagram, which test case below covers every valid transition?
a. SS-S1-S2-S4-S1-S3-ES
b. SS-S1-S2-S3-S4-S3-S4-ES
c. SS-S1-S2-S4-S1-S3-S4-S1-S3-ES
d. SS-S1-S4-S2-S1-S3-ES
Question 36 A test plan included the following clauses among the exit criteria:
• System test shall continue until all significant product risks have been covered to the extent specified in the product risk analysis document.
• System test shall continue until no must-fix defects remain against any significant product risks speci fied in the product risk analysis document.
During test execution, the test team detects 430 must-fix defects prior to release and all must-fix defects are resolved. After release, the customers find 212 new defects, none of which were detected during testing. This means that only 67% of the important defects were found prior to release, a percentage which is well below average in your industry. You are asked to find the root cause for the high number of field failures. Consider the following list of explanations:
I Not all the tests planned for the significant product risks were executed.
II The organization has unrealistic expectations of the percentage of defects that testing can find.
III A version-control issue has resulted in the release of a version of the software that was used during early testing.
IVThe product risk analysis failed to identify all the important risks from a customer point of view.
V The product risk analysis was not updated during the project as new information became available.
Which of the following statements indicate which explanations are possible root causes?
a. II, III and IV are possible explanations, but I and V are not possible.
b. All five are possible explanations.
c. I, IV and V are possible explanations, but II and III are not possible.
d. Ill, IV and V are possible explanations, but I and II are not possible.
Question 37 What is the most important factor for successful performance of reviews?
a. A separate scribe during the logging meeting
b. Trained participants and review leaders
c. The availability of tools to support the review process
d. A reviewed test plan
Question 38 Consider the following statements about maintenance testing:
I It requires both re-test and regression test and may require additional new tests.
II It is testing to show how easy it will be to maintain the system.
III It is difficult to scope and therefore needs careful risk and impact analysis.
IVIt need not be done for emergency bug fixes. Which of the statements are true?
a. I and III
b. I and IV
c. II and III
d. II and IV
Question 39 Which two specification-based testing techniques are most closely related to each other?
a. Decision tables and state transition testing
b. Equivalence partitioning and state transition testing
c. Decision tables and boundary value analysis
d. Equivalence partitioning and boundary value analysis
Question 40 Which of the following is an advantage of independent testing?
a. Independent testers don't have to spend time communicating with the project team.
b. Programmers can stop worrying about the quality of their work and focus on producing more code.
c. The others on a project can pressure the independent testers to accelerate testing at the end of the schedule.
d. Independent testers sometimes question the assumptions behind requirements, designs and implementations.
ANSWERS TO SAMPLE EXAM QUESTIONS
This section contains the answers and the learning objectives for the sample questions in each chapter and for the full mock paper in Chapter 7.
If you get any of the questions wrong or if you weren't sure about the answer, then the learning objective tells you which part of the Syllabus to go back to in order to help you understand why the correct answer is the right one. The learn-ing objectives are listed at the beginning of each section. For example, if you got Question 4 in Chapter 1 wrong, then go to Section 1.2 and read the first learn-ing objective. Then re-read the part of the chapter that deals with that topic.
Sunday, April 12, 2015
Subscribe to:
Posts (Atom)