1971 – 1985: Emphasize on Implementation and Single Program:

Beginning from the mid 1960s to the mid 1980s, the whole software engineering research community

shifted it paradigms to the program-in-the-small stage, and then started the program-in-the-large stage.

The main changes this migration brought to software development were that the characteristic

problems changed from small programs, to larger programs and algorithms, and were on the way to

developing more complex problems. In response to this significant change, researches on testing

techniques began their prosperity.

On the structural side, in 1975, 1976, 1980, and 1985, four significant papers were published, each

proposing a very important structural testing strategy, and all of which were adopted as the classic

criteria for later researches followed them. From Figure 3 we can see that almost all the important

13

theoretical structural testing researches appeared in this period. Although the whole software

engineering community was facing the challenge of switching the gear of developing from comparably

simple programs to complex large systems, it took time for testing community to react to the change,

specifically, in approximately 5 years.

From the figure we also find that only one significant result for functional testing appeared in this

period. The reason is obvious. Functional testing is based on requirements and has consisted merely

of heuristic criteria. It is difficult to determine when and if such criteria are satisfied without being

able to exp ress the requirements in an efficient, rigorous, unambiguous way. This was in part the

motivation for developing implementation-based testing techniques; they have the advantage that their

application can be automated and their satisfaction determined. Fortunately the research appeared

during this period set up a very good tone of successive researches, since it moved emphasis from the

simple input/output specifications that testers often used in this period to a higher level – the design of

the system.

In this period, how to test a “program”, instead of a “system”, still drew the attention of researchers

and practitioners. However the whole software engineering had begun to get ready for moving from

the stage of programming-in-the-large to a higher level.

1986 – current: Emphasize on Specification and System:

As software become more and more pervasive, the engineering for this area experienced the shift from

programming-in-the-large to programming-in-the-world, starting from the mid 1980s. The

characteristic problems changed from algorithms, to system structures, and component interfaces.

Systems have been specified in more complex ways. Studies in software architecture and formal

methods have brought a lot of facilities as well as inspiration to the way people specifying their

systems. Based on these studies, software system now can be specified in more rigorous,

understandable, automatable ways, which has brought great chances to improve functional testing

techniques. Meanwhile, software development is no longer limited to standalone systems, in reality,

there have been more and more needs to develop distributed, object-oriented, and component based

systems. The researchers in testing community have responded this trend and move their emphasis

accordingly.

Starting from the late 1980s, many researches have made use of the achievements of formal methods

and logical analysis. There is still limitation in the specification capabilities so that researchers have

been calling for better specification methods to improve their results. Both functional and structural

testing techniques have benefited from the enhancement of software specification technologies.

The widespread developing and using of object-oriented technologies, COTS software and componentbased

systems has brought a great density of testing researches on these kinds of systems. The earliest

OO testing studies appeared in the early 1990s. Most of them use traditional functional and/or

structural techniques on the components, i.e. classes and so on. Researchers have proposed new

problems and solutions on testing the connections and inheritances among components. Both

structural and functional techniques are hired in their approaches, and it has proven to be an effective

method to integrate the two techniques for testing complex systems.