Bulletin

Issue: May 2004

Research in Action

Software testing There's an easier way

A La Trobe University software engineer believes she may have resolved a problem that has plagued the computer industry since its inception.

PhD student and sessional lecturer in Computer Science and Computer Technology, Ms Tafline Murnane, has devised a new model to describe software testing methods.

She believes the model will change the way software testing is taught and may change the way software testing methods are described in text books on testing.

As part of her research for her PhD thesis, Ms Murnane has created the model to make software testing methods easier to learn, use and compare against each other.

She explains that many text books used in software testing and reliability courses were written 20 years ago and are still used because the methods have not changed significantly.

'We did not realise there were so many 'holes' in the text books where they describe testing methods. Because of this, the methods can be difficult to learn as a number of steps tend to be left out of each text book description of the methods.'

Ms Murnane calls her new model the TCGM Model, which stands for the Test Case Generation Model, as it relates to the generation of test cases.

'Currently, when software testing methods are described in text books, each one is described differently. I searched the literature and was unable to find anything relating to how testing methods could be described in ways that make them easier for students to learn.

'Of eight testing methods I have researched, all are described differently and this makes it difficult not only for students to learn, but also for practitioners using the methods in industry. The TCGM Model gives all testing methods the same look and feel, which will help teachers, students, researchers and industry practitioners.

'If you look at the ingredients lists of packet or processed food, they are all structured the same way. Because of this you don't have to learn a new method of reading the lists for each different food. That is not the case with software testing methods, as you have to learn a different way of representing the methods for each new method you learn. To overcome this, I devised this standard model for describing software testing methods.

'In addition, the model also makes testing more flexible. Essentially, software testing methods consists of different sets of rules for selecting test data. The model describes the test selection rules independent of the methods they were derived from, so that if an industry practitioner wants to generate a test case, instead of applying an entire testing method, they can use just those rules that apply to their particular problem domain.

'The model is a more practical alternative to current testing methods, because each test can be customised to the specific needs of the tester or organi-sation,' Ms Murnane said.

Professor Karl Reed and Dr Richard Hall are supervising Ms Murnane who is doing her PhD at La Trobe on a Victorian State Government ITC Scholarship and a scholarship from the Department of Computer Science and Computer Engineering. She worked in industry as an analysist programmer after completing her degree in Computer Systems Engineering at La Trobe in 1999.•

back to top

back contents next