CSE565 Fall 2010 | Week 05:
Structure-Based Testing (White Box Testing)


CSE565: Software Validation, Verification and Testing
Department of Computer Science and Engineering Ira A. Fulton School of Engineering
Arizona State University

 

This are my notes for the eighth class session:

1. Structure-Based Testing (White Box Testing)

  • A white box testing methods gets its name due the fact it see the code. Unlike black box testing methods that are all based only in requirements.
  • Tend to be techniques to be applied into lower level (small units of code) unit, services, components.
  • Not for entire systems
  • Not fot Integration or Qualification Level of Testing

There are two categories of structure-based Testing (white box testing): static techniques and dynamic techniques.

1.1. Static Techniques

Analysis of the code. Example: symbolic execution.

1.2. Dynamic Techniques

Execute the code: run for test. There are two types of dynamic techniques:

  • Structure or control flow.
    • Statement coverage
    • Decision coverage
    • Decision-Condition Coverage
    • Multiple decision Coverage
  • Data flow

2. Statement Coverage

Develop test cases such that every statement is execute at least once.

  • Example:

    if a <10 or b>5
    (1) then x<-50 (2) else x <-0; if w = 5 or y>0
    (3) then z<-48 (4) else z<-5;

  • Flow Chart = Control Flow Diagram.
    Test predicate (diamond)
    Statement (oval)
  • 4 statement in the previous code.
  • Path Testing to execute every Path into the code!
    Testing strategy execute all path into the code

  • Loops are going to make the numbers of paths bigger. So path coverage is complicate.
  • Values of (a, b, w, y) to execute all statements (with 2 cases we do it)!

    a = 0, b=0, w=0, y=0 (execute 1 and 4)
    a=10, b=4, w=5, y = don’t care (execute 2 and 3)

    Whit these two tests cases we are covering all the statements.

  • However, this kind of testing is the minimum need.
    FAA D0178B standard / best practices:
    1. statements 100% statement coverage!
  • Compilers help us to do that. By example the PROFILER measure the time of each statement.

3. Decision (branch) Coverage

  • Develop test cases such that each branch (decision point) is executed at least once.
  • Decision point -> conditional statement = test predicates = diamonds
  • EJEMPLO UN MAPA CUADRICULAR CON CALLES Y CADA INTERSECION ES DECIDIR UNA DIRECCION
  • Even If I use structured programming (no GOTO) then if I have NOT 100% statement coverage then I have 100% test predicated.

    s1
    if x<10 then s2 s3

    x =5
    I have 100% statement coverage
    But NOT 100% decision coverage, false case is MISSING!
  • Statement Coverage Do Not Imply Decision Coverage!
  • Decision Coverage satisfies Statement but Statement not necessarily satisfies Decision!
    So Software must cover 100% Decision Coverage

4. Decision-Condition Coverage

  • Develop test such that each condition in a decision takes on all possible outcomes and each decision takes on all possible outcomes
  • This is like Decision++ (the second part of the definition, is Decision Coverage)

    if x<10 or y>50 and z!=0 or flag= T and status = NIL
    then S1
    else S2

    2 test to statement coverage
    2 test to decision coverage

    But that not enough for Decision-Condition Coverage.
    We need to open the test predicate and look inside: there are 5 conditions!

    For each condition we need to test their TRUE and FALSE cases

    We need 2 cases. All false and All True.

5. Multiple Condition

  • Develop test to execute all combinations of conditions within a decision.
    This is the most powefull.

6. Example, Test Binary Search Algorithm)

(1) start <-1; (2) end <-num; (3) found <- false; (4) while starts <= end and not found //(C1) and (C2)
(5) middle <-(start = end)/2 (6) if key > table[middle] //(C3)
(7) then start <- middle +1 (8) else if key = table[middle] //(c4)
(9) then found <-T (10) LOC <- middle; (11) else end <- middle -1

We have 4 conditions, and we have 11 statements (each row). We have three decision points and we have one case with multiple (2) conditions.

if4
T
F

if6
T
F

if8
T
F

For multiple-condition coverage C1 y C2 make combinations of T y F

For conditions C3 y C4 simply try one T and one F of each one

Example of Execution

10 20 30 40 50 60 70 Key=55

1
2
3
4 with T, T
5
6 T
7

4 done before T, T
5
6 F

8 with F

11

4 con T, T
5
6 T
7

4 con F, T

  • Multiple-condition Coverage: 100% NO
  • Decision-condition Coverage: 100% NO
  • Decision Coverage: NO
  • Statement coverage: NO

Now execute this:

10 20 30 40 50 60 70 Key = 40

1
2
3
4 with T, T
5
6 with F

8 with T

9
10

4 with T,F

Coverage is cumulative

4. Multiple-condition Coverage NOT 100% (75%)
3. Decision-condition Coverage 100%
2. Decision Coverage 100%
1. Statement Coverage 100%