How does a DU path segment play a role in data flow testing?

Whenever you would have came across the topic of data flow testing, you surely would have heard about the term “du path segment” but still not familiar with it! This article if focussed up on the du path segments and what role it has got to play in the data flow testing. 

We will discuss the du path segments under the context of data flow testing and not as a separate topic so that it becomes easy for you to understand. 

The whole process of data flow testing is guided by a control flow graph that apart from just guiding the testing process also helps in rooting out the anomalies present in the data flow. With all the anomalies being already discovered one can now design better path selection strategies taking in to consideration these data flow anomalies. 

There are nine possible anomalies combinations as mentioned below:

1. dd: harmless but suspicious
2. dk: might be a bug
3. du: a normal case
4. kd: a normal situation
5. kk: harmless but might be containing bugs
6. ku: a bug or error
7. ud: not a bug because of re- assignment
8. uk: a normal situation
9. uu: a normal situation

For data flow testing some data object states and usage have been defined as mentioned below:

1.      Defined, initialized, created à d

2.      Killed, undefined, unreleased àk

3.      Used for:

(a)    Calculations à c

(b)   Predictions à p

Terminology associated with Data Flow Testing

Almost all the strategies that are implemented for the data flow testing are structural in nature. There are certain terminologies associated with the data flow testing as stated below:

1. Definition clear path segment
2. Loop free path segment
3. Simple path segment and lastly
4. Du path

What is a DU path Segment?

- DU path segment can be defined as a path segment which is simple and definition clear if and only if the last link or node of the path has a use of the variable x.

Let us take an example to make the concept of du path segment clearer. 

- Suppose a du path for a variable X exists between two nodes namely A and B such that the last link between the two nodes consists of a computational use of the variable X. 

- This path is definition clear and simple. 

- If there exists a node C  at the last but one position that is the path is having a predicate use and the path from the node A to node C is definition clear and does not contain any loop. 

- Several strategies have been defined for carrying out the data flow testing like:

1. ADUP or all du paths strategy
2. AU or all uses strategy
3. APU+ C or all p uses/ some c uses strategy
4. ACU +P or all c uses/ some p uses strategy
5. AD or all definitions strategy
6. APU or all predicate uses strategy
7. ACU or all computational uses strategy

Strategy for DU Path Strategy

We shall describe in detail here only the ADUP or all du paths strategy. 

- This strategy is considered to be one of the most reliable and strongest data flow testing strategy. 

- It involves the use or exercising of all the du paths included in the definition of the variables that have defined to every use of the definition.

- All the du paths suppose to be a strong criterion for testing but it does not involve as many tests as it seems.

- Simultaneously many criterion are satisfied by one test for several definitions and uses of the variables.

How does a definition use association play a role in data flow testing?

Definition use association is one of the terms that appear at the scene of data flow testing and quite many of us are unaware of it. This article is all about the concepts of the definition use associations and what role does they have got to play in the data flow testing. 

The definition use association forms quite an important part of the data flow testing. Let us see how! 

First we are going to discuss some concepts of the data flow testing in regard with the definition use associations and then we will discuss the role of the definition use associations in the data flow testing. 

About Data Flow Testing

- A control flow graph is an important tool that is used by the data flow testing so that the anomalies related to the data can be explored. 

- A proper path selection strategy is what that is required for detection of such anomalies. 

- The path strategy which is to be used can be decided on the basis of the data flow anomalies discovered earlier. 

- Data flow testing is nothing but a family of path testing strategies through the control of a software program or application. 

- The path testing is required so that the sequence of possible events associated with the objects’ status can be explored.

- It is necessary that you keep the number of paths sufficient and sensible so that no object is left without initialization and without being used at least once in its life time without carrying out unnecessary testing. 

- Data flow testing is comprised of two types of anomaly detection namely:

1. Static analysis: It is carried out on the program code without its actual execution. It involves finding syntax errors.

2. Dynamic analysis: It is carried out on a program while it is in a state of execution. It involves finding the logical errors.

- The data objects have been categorized in to several categories so as to make data flow testing much easy:

1. Defined, created, initialized (d)
2. Killed, undefined, released (k)
3. Used (u): in calculations (c)
4. In predicates (p)

Anomalies discovered by the static analysis are meant to be handled by the compiler itself. But the static analysis and the dynamic analysis do not suffice altogether. A rigorous path testing is required. 

About Definition Use Associations

- The definition use associations or the “du segments” are the path segments whose last links have a use of variable X and are simple and definition clear. 

- Typically  a definition use association is a combination of triple elements (x, d, u) where:

1. X is the variable
2. D is the node consisting of a definition of variable x
3. U is either a predicate node or a statement depending up on the case and consists of a use of x.

- A sub path from d to u is also included in the flow graph with no definition of variable x occurring in between the d and u. 

- Below mentioned are some examples of the def use associations:

1. (x, 3, 4)
2. (x, 1, 4)
3. (y, 2, (4, t))
4. (z, 2, (3, t)) etc.

- Some of the most common data flow testing strategies are:

1. All uses (AU)
2. All DU paths (ADUP) and many more.

- First advice for effective data flow testing would be to resolve all the data flow anomalies discovered above. 

- Carrying out data flow operations on the same variable and within the same routines can also reap you good results. 

- It is advisable to use defined types and strong typing wherever it is possible in the program. 

How does a definition clear path play a role in data flow testing?

Definition clear path is a quite less heard term! This article is focussed up on the concept of definition clear path and what role do it plays in the data flow testing. First let us define what is a definition clear path in actual. 

"A definition clear path as it can made out from the term itself that it is a path through which other variables cannot be defined or through which other variable definitions cannot be made."

To make the meaning of definition clear path clearer we shall look up to an example:

- Suppose X be a variable declared or appearing in a software program or procedure. 

- Suppose there is a path $which\; do\; not\; contain\; any\; nodes\; with\; definition\; of\; the\; variable\; X.$

- Such a path not containing any variable definitions has been termed as a definition clear path. 

We can now define this definition clear path as a path between the two nodes namely A and B, with X being defined in A and an use in node B and there exists no other definition of variable X between the two nodes present in the path. 

Let us see another example to explore another type of definition clear path that can exist. 

- Suppose the above same variable X be defined at a node A along with an use defined at the another node B.

- Suppose the path formed by these two nodes A and B does not appears in the sub path, then such a path is also defined as a definition clear path for the X variable defined by the nodes A and B if the variable X is not defined in the sub path. 

- There is another common name for the definition clear path which is “def- clear path”. 

Now let us talk about the role that the definition clear path plays in the data flow testing. Actually in the data flow testing, there are three types of coverage that have to be provided namely:

1. Statement coverage
2. Branch coverage and lastly
3. Path coverage

Basically problems are faced with the path selection process. A definition of the variable X reaches a use if and only if there exists a sub path such that the sub path is a definition clear path with respect to the variable X. The path selection in the data flow testing is based up on the two criteria:

1. Rapps and Weyuker criteria: Under these criteria the definition clear sub paths from definitions to uses are listed.
2. Laski and Korel criteria: Under these criteria the various combinations that reach uses at a node via some sub path are listed.

How does Definition Clear Path play a role in Data Flow Testing?

- Definition clear paths have been known to make remarkable improvements in the control flow techniques for data flow testing.

- A rational is obtained for which there is a need to take in to consideration all the combinations of the sub paths. 

- The “all uses” is the most commonly preferred criteria.

- There are some paths in a program that are infeasible and it is these paths that pose a big problem in the data flow testing. 

- The path testing strategies are based up on the data flow anomalies. 

- Enough paths are required to be tested so that it is ensured that every object in the program has been initialized before use and have been used at least once during the program execution. 

- For a complete data flow testing it is required that definition clear paths are executed by the test cases from each node that contains a defined variable.

What are different data flow testing strategies?

Data flow testing is quite important since you do not want any unreasonable things happen to your data objects which in turn can deviate the whole control flow of your program from the right track. To make a sensible data flow testing you need to use sensible and reliable data flow testing strategies.

This article is all about such data flow testing strategies. 

There are two types of machines that are used by the data flow as mentioned below:

1. Von Neumann machine architecture
2. Multi- instruction, multi- data machines architecture (MIMD)

- Before carrying out the data flow testing, it is good to assume a bug which causes problem in the control flow of the program. 

- It is not compulsory to use the typical data flow graphs, annotated ordinary control graphs can also be used for guiding the data flow testing process. 

- Data flow graph depicts all the directed links and nodes involved in the data flow.

- All the strategies for data flow testing that we are going to discuss are structural in nature and also focus up on the actions taking place on the data objects rather then just focussing on the connectivity of the software program.

Requirements of Data flow testing Strategy

- Data flow link weights are the first requirement of any data flow testing strategy. 

- All these strategies are based up on the selection of the path segments that very well satisfy at least few of the data flow characteristics common to all the data objects. 

- A data flow testing strategy is weaker than another strategy Y if all the test cases present in Y are not included in the X. Then Y is said to be a stronger strategy. 

Important Terminologies

Let us take a look at some important terminologies before moving on to the strategies:

1. Definition clear path segment: It is a path defined with respect to a variable X that consists of various links such that the X is defined only on the first link.
2. Simple path segment: In such a path one of the two nodes are visited twice.
3. Loop free path segment: This path is contrary to the simple path segment in the way that in this path every node is visited once for the maximum.
4. Du path segment: It is a path that is simple and definition clear since its last link consists of a computational use of variable X.

Different Strategies for Data Flow Testing

Below described are the different strategies for the data flow testing:

1. ADUP or all DU paths: This strategy is considered to be the strongest among all the data flow testing strategies. It takes in to account all the du paths that occur in the definitions of all the variables to their every use. This strategy is a strong data flow testing criteria also. Another advantage of this strategy is that one of its tests can satisfy many definitions at a time.
2. AU or all uses strategy: Under this strategy at least one of the definition clear paths from all the definitions of a variable has to be tested or exercised under a test. The task or burden of testing is actually reduced here i.e., the path coverage is cut down to branch coverage.
3. APU + C or all p uses/ some c uses strategy: This strategy covers up at least one definition free path to every predicate use for every definition of the function. If this is not able to over up all the definitions of the variable, then it is recommended that computational use test cases are exercised.
4. ACU + P or all c uses/ some p uses strategy: This strategy is just the opposite of the above strategy.
5. AD or all definitions strategy: It covers only the definition of the variable. 

How does a loop free path segment play a role in data flow testing?

The loop free path segments form a very important terminology in the path of data flow testing. But many of us are not well familiar with the concept of loop free path segments and the role that they have got to play in the data flow testing or path testing. In this article we have tried to explain in the easiest way possible the concept of the loop free path and the role it plays in the data flow testing. 

Before taking up the topic of the loop free path segment and its role in data flow testing we shall discuss a little about the data flow testing. 

The control flow graph is the best tool that the data flow testing can use in exploring all the weird or unreasonable things that can affect the data objects. These weird and unreasonable things are nothing but the anomalies.

Till now nine types of anomalies have been defined as mentioned below:

1. dd: harmless but suspicious
2. dk: might be a bug
3. du: a normal case
4. kd: a normal situation
5. kk: harmless but might be containing bugs
6. ku: a bug or error
7. ud: not a bug because of re- assignment
8. uk: a normal situation
9. uu: a normal situation

 - If these anomalies are taken in to consideration, one can develop very effective and reliable path selection strategies which can be then used in filling the gaps that are present in between the branch testing and the complete path testing. 

- The strategies that are followed for carrying out a data flow test are based up on the selection of the paths via the flow of control of the software system or application.

- These path selection strategies are quite useful when it comes to exploring the sequences of the events that are in a way related to the status of the data objects. 

- The paths are so selected that they cover up all the objects, i.e. they ensure the initialization of each and every data object before it is used in the program and also that they are used for a minimum of one time. 

Categories of Data Objects

- The data objects have been categorized in to three different categories for making the path selection process easier:

1. Defined, created, initialized (d)
2. Killed, undefined, released (k)
3. Used:

(a)    In calculations (c)

(b)   In predicates (p)

- An object is said to be defined whenever it has an occurrence in a data declaration or is assigned with a new value or is dynamically allocated. 

- On the other hand an object is said to be used whenever it becomes a part of a predicate or a calculation. - The anomaly detection process relies heavily on the following two anomaly detection techniques:

1. Static anomaly detection (responsible for syntax errors) and
2. Dynamic anomaly detection (responsible for logical errors).

What are Loop Free Path Segments

- Now coming to the loop free path segments, this is a terminology that is usually used under the context of the data flow modelling. 

- Loop free path segments are discovered using the control flow graph.

- The loop free path segments are basically a derivative of the simple path segments.

- It depends on the simple path segment that whether or not it is a loop free path segment also. 

- If the simple path segment consisting of two nodes A and B is having loop in both the nodes, then it cannot be called as a loop free path. 

- Loop free paths are the simple paths segments consisting of loop only in one of the either nodes.

How does a simple path segment play a role in data flow testing?

Whenever you have discussed about the data flow testing you must have came across the term simple path segments while discussing about the strategies for data flow testing. Many of us are not quite clear with the concept of the simple path segments and what role have they got to play in the data flow testing. This article is all about the simple path segments and the role that they have got to play in the data flow testing. 

First we shall brief up ourselves with the concepts of the data flow testing before moving on to the topic of the simple path segments and their role. 

What is Data Flow Testing

- Data flow testing includes all those strategies that have been based up on the selection of the paths via the control flow of the program for discovering the sequence in which the events related to the object’s status take place.

- A primary bug assumed during the data flow testing is that though the control flow is generally correct, there is some fault with the software system or application since the data objects are not available when they are supposed to be or weird things happen to the data objects. 

- Even if some problem is found to preside in the control flow of the program it is initially detected by the data flow analysis. 

- One of the most aiding tool in the data flow testing are the control flow graphs which are the graphs consisting of directed links and nodes. 

- The objective of the data flow testing is to discover the deviations in the data flow. Three types of data objects have been defined namely:

1. Killed or undefined
2. Defined
3. Usage

And some nine kinds of anomalies have been defined:

1. dd: harmless but suspicious
2. dk: might be a bug
3. du: a normal case
4. kd: a normal situation
5. kk: harmless but might be containing bugs
6. ku: a bug or error
7. ud: not a bug because of re- assignment
8. uk: a normal situation
9. uu: a normal situation

These anomalies are detected by the means of two anomaly detection techniques namely:

1. static anomaly detection technique and
2. dynamic anomaly detection technique

All the strategies involved in the process of data flow testing are structural. Data flow testing and path testing strategies have so many things in common. But one of difference between them is made by what one takes in to account for testing. Path and data flow testing both are emphasized up on the raw connectivity of the graph but in addition to this the data flow testing also focuses up on what happens to the data objects. There are so many terminologies associated with the data flow testing and simple path segment is one of them. The others are:

1. definition clear path segment
2. loop free path segment
3. du path segment

What is Simple Path Segment

We shall now define what a simple path segment is! 

- Any path in which the same node is visited twice at the most, such a path is called a simple path segment. 

- One can easily make out why a simple path segment is called so! 

- It is called so because it does not consists of loops in both the nodes.

- Only one node holds the loop. 

- One of the problems that are faced by the testers is of finding the simple paths. 

- This problem can be overcome by following a lower bound max- flow approach. 

- The simple path segments though being, are important in the data flow testing just like all the other path segments. 

Data flow testing is a white box testing technique - Explain?

A program is said to be in active state whenever there is some data flow in the program. Without having the data flowing around the whole program, it would not have been possible for the software systems or application to do any thing.

Hence, we conclude that data flow is an extremely important aspect of any program since it is what that keeps a program going on. This data flow also needs to be tested like any other aspect of the software system or application and therefore, this whole article is dedicated to the cause of the data flow testing.

What is Data Flow Testing?

- Data flow testing technique has been categorized under the white box testing techniques since the tester needs to have an in depth knowledge of the whole software system or application.

- Data flow testing cannot be carried out without a control flow graph since without that graph the data flow testing won’t be able to explore any of the unreasonable or unexpected things i.e., anomalies that can influence the data of the software system or application.

- Taking these anomalies in to consideration, it helps in defining the strategies for the selection of the test paths that play a great role in filling up the gaps between the branch testing or statement testing and the complete path testing.

- Data flow testing implements a whole lot of testing strategies chosen in the above mentioned way for exploring the events regarding the use of the data that occurs in a sequential way.

- It is a way determining that whether or not every data object has been initialized before it used and whether or not all the data objects are used at least once during the execution of the program.

Classification of Data types
The data objects have been classified in to various types based up on their use:

- Defined, created and initialized data objects denoted by d.
- Killed, undefined and released data objects denoted by k.
- Used data objects in predicates, calculations etc, denoted by u.

Critical Elements for Data Flow Testing

- The critical elements for the data flow testing are the arrays and the pointers.

- These elements should not be under estimated since they may fail to include some DU pairs and also they should not be over estimated since then unfeasible test obligations might be introduced.

- The under estimation is preferable over the over estimation since over estimation is causes more expense to the organization.

- Data flow testing is also aimed at distinguishing between the important and not so important paths.

- During the data flow testing many a times pragmatic compromises are needed to make since there exist so many unpredictable properties and exponential blow up of the paths.

Anomaly Detection under Data Flow Testing

There are various types of anomaly detection that are carried under the data flow testing:

1. Static anomaly detection
This analysis is carried out on the source code of the software program without the actual execution.

2. Dynamic anomaly detection
This is just the opposite of the static testing i.e., it is carried out on a running program.

3. Anomaly detection via compilers
Such detection are possible due to the static analysis. Certain compilers like the optimizing compilers can even detect the dead variables. The static analysis itself is incapable of detecting the dead variables since they are unreachable and thus unsolvable in the general case.

Other factors:
There are several other factors that play a great role in the data flow testing and they are:
1. Data flow modelling based on control flow graph
2. Simple path segments
3. Loop free path segments
4. DU path segments
5. Def – use associations
6. Definition clear paths
7. Data flow testing strategies

Control Structure Testing - Condition Testing, Data Flow Testing, Loop Testing

Control structure testing is a group of white-box testing methods.
CONDITION TESTING
- It is a test case design method.
- It works on logical conditions in program module.
- It involves testing of both relational expressions and arithmetic expressions.
- If a condition is incorrect, then at least one component of the condition is incorrect.
- Types of errors in condition testing are boolean operator errors, boolean variable errors, boolean parenthesis errors, relational operator errors, and arithmetic expression errors.
- Simple condition: Boolean variable or relational expression, possibly proceeded by a NOT operator.
- Compound condition: It is composed of two or more simple conditions, Boolean operators and parentheses.
- Boolean expression: It is a condition without Relational expressions.

DATA FLOW TESTING
- Data flow testing method is effective for error protection because it is based on the relationship between statements in the program according to the definition and uses of variables.
- Test paths are selected according to the location of definitions and uses of variables in the program.
- It is unrealistic to assume that data flow testing will be used extensively when testing a large system, However, it can be used in a targeted fashion for areas of software that are suspect.

LOOP TESTING
- Loop testing method concentrates on validity of the loop structures.
- Loops are fundamental to many algorithms and need thorough testing.
- Loops can be defined as simple, concatenated, nested, and unstructured.
- In simple loops, test cases that can be applied are skip loop entirely, only one or two passes through loop, m passes through loop where m is than n, (n-1), n, and (n+1) passes through the loop where n is the maximum number of allowed passes.
- In nested loops, start with inner loop, set all other loops to minimum values, conduct simple loop testing on inner loop, work outwards and continue until all loops tested.
- In concatenated loops, if loops are independent, use simple loop testing. If dependent, treat as nested loops.
- In unstructured loops, redesign the class of loops.

Control Structure Testing : Data Flow Testing

Data Flow Testing is a technique which is used effectively alongside Control Flow Testing. It is another type of white-box testing which looks at how data moves within a program. Data flow occurs when variables are declared and then accessed and changed as the program progresses.
A "definition-clear path" is a path which involves no changes of the variable being tested. For a statement with S as its statement number,

DEF(S) = {X| statement S contains a definition of X}
USE(S) = {X| statement S contains a use of X}

If statement S is an if or loop statement, its DEF set is left empty and its USE set is founded on the condition of statement S. The definition of a variable X at statement S is live at statement S’ if there exists a path from statement S to S’ which does not contain any condition of X.

A definition-use chain (or DU chain) of variable X is of the type [X,S,S’] where S and S’ are statement numbers, X is in DEF(S), USE(S’), and the definition of X in statement S is live at statement S’.

One basic data flow testing strategy is that each DU chain be covered at least once. Data flow testing strategies are helpful for choosing test paths of a program including nested if and loop statements.