Effectiveness of software testing techniques in enterprise: a case study

Testing shows presence of defects
: that was discussed in previous paragraphs as the possibility of 
"zero-defect" software. Is was concluded that this theoretical assumption cannot be applied in practice. 
Even after testing is completed it cannot be stated that the product reached the effect "zero-defect". 
Testing always reduces the number of undiscovered defects, but there are always remained issues that 
could be detected in the last phase of SDLC as we discussed before (the rise from error to failure) or 
even more later if the undiscovered defect affects only one component of software which is used 
rarely. Moreover, even if no defects are found, it is not a proof of correctness as well. 
2) 
Exhaustive testing is impossible
: testing everything including all combinations of inputs and 
preconditions is not possible because of time constraints, the vast of data used in preconditions or for 
inputs, limited human capabilities to overcome the full testing (all combinations of inputs and 
preconditions). Myers et al. (2011) follows the idea of exhaustive testing and adds some implications 
of this: it is impossible to test a software to guarantee that it is error free (or "zero-defect" as it was 
discussed in the first principle) and a fundamental consideration in software testing is one of 
economics. Thus, instead of doing the exhaustive testing, the risk and priorities are suggested to use to 
distribute testing efforts. As in the researches (Elbaum et al., 2002; Rothermel et al., 1999; Srivastava, 
2008), test case prioritization can be used for assisting with regression testing by setting priorities for 
test cases and performing testing from the test with highest priority (lowest priority test can be skipped 

---

24
in case of tight schedule by taking some risk). These techniques will be discussed more detailed in 
further chapters. Therefore, accessing and managing risk take an important role in QA activities in any 
project.
3) 
Early testing
: software testing activities should start as early as possible in order to avoid costly 
failures in SDLC. As it was concluded before - all mistakes made in early stages of SDLC could rise to 
failures; thus this incident affects the quality of software.
4)