5.3 Testing Process Models:

The third reference model is the major testing goals model given by Gelperin and Hetzel in [6]. We have

discussed this in detail in our previous section of testing concept evolution. Understanding how the goals

of testing has changed helps us a lot to understand the intellectual history of testing technique research,

how new ideas are based on old ideas from the same area or on the studies from other areas.

The Redwine/Riddle model provides benchmarks for judging where a research stands in the maturation

process, and how a process can migrate to the next one. Shaw’s model sets up the global background

where studies of technology maturation can be put into and compared with the rest of the world of software

engineering. Gelperin and Hetzel’s model is specific in testing so that the objectives of a testing technique

research are made clear. Each model assists us to understand the problem from a different point of view,

but none of them alone fully explains the unique phenomena in the technology history of test techniques.

We put our study in the framework built with these models, and give our own viewpoints.

5.4 The Major Stages of Research & Development Trends

Generally, we see three major stages of the research and development of testing techniques, each with a

different trend. By trend we mean the how mainstream of research and development activities find the

problems to solve, and how they solve the problems. As is shown in the column of “Technology

Evolution” in Figure 3, testing technique technologies, thus the ways of selecting test data have developed

from ad hoc, experienced implementation-based phase, and is focusing on specification-based now.

1950 – 1970: Ad Hoc:

From Figure 3 we can see that between the years 1950 and 1970, there were few research results on

testing techniques except for the conceptual ideas of testing goals. It’s possible that research results

before 1970 are too old to be in the reach of current bibliography collections. To avoid being

influenced by this factor, we looked at many testing survey papers in the 1980s, which should have had

the “ancient” studies in hand by the time they performed their study. We suppose their surveys at least

addressed the most important technical contributes before their time, and we can build our research for

the decades before 1970 on theirs.

Based on above assumption, we define the period between 1950 and 1970 as being ad hoc. During this

period, major research interest focuses on the goal of testing, and there are quite a few discussions on

how to evaluate if a test is good. Meanwhile, testing had become gradually independent from part of

debugging activities, to a necessary way to demonstrate that a program satisfies its requirements, as is

seen in the GH88 model. At the same time, if we look from Shaw’s view, we can see that the whole

world of software engineering was in its programming-in-any-which-way stage. It’s very natural that

testing stayed in its ad hoc stage, where test data is selected randomly and in an unorganized,

undirected way.