Derek Yates

The collateral coverage of data flow criteria when branch testing

Abstract When exercising program code with test data in an attempt to satisfy a given testing criterion, there will be a concurrent accrual of coverage in respect of other testing criteria. Knowledge of the extent of such ‘collateral coverage’ can be used to advantage both in providing better estimates of the overheads entailed by the overall testing exercise, and in helping to determine an optimal sequence for the application of a set of testing methods. In this paper, the results deriving from a set of experiments are reported. The aim of the experiments was to investigate the extent of the collateral coverage that is achieved in respect of the data-flow testing criteria when branch testing is undertaken.

An objective comparison of the cost effectiveness of three testing methods

Branch testing is a well established method for exercising software. JJ-path testing, whilst employed by some practitioners, is less popular, and the testing of JJ-pairs finds few adherents. In this paper an objective, practical study of the cost-effectiveness of these three testing methods is reported. The effectiveness of each method is assessed, in the presence of infeasible paths, not only on its ability to cover the specific structural element of code that it targets, but also on its ability to cover the structural elements targeted by the other two methods - the collateral coverage it achieves. The assessment is based on the results derived from experiments in which each of the three methods is applied to 35 units of program code.

Inclusion, subsumption, JJ-paths, and structured path testing: a Redress

A liquid crystal display panel has a first substrate and a second substrate composed of a transparent material and disposed opposite each other across a gap, display electrodes provided on one of the opposed surfaces of the first and second substrates, transparent counter electrodes provided on the other opposed surface, a region between each display electrode and an opposed counter electrode constituting a pixel region, a liquid crystal layer filled in the gap between the first substrate and the second substrate, and a color filter or a reflecting film provided at each pixel region. The color filter or the reflecting film at each pixel region is disposed as divided into multiple segments and light transmitting portions are provided around the color filter or reflecting film. Otherwise each color filter or reflecting film is partially provided with multiple openings or other portions with high light transmittance.

Editorial: Uncertainty and chaos

At the beginning of the 19th century, Laplace argued that, given complete knowledge of the state of the universe at a given time, everything that then occurred should be predictable using some set of scientific laws. This ‘determinism’ gained favour with the scientific community, and held sway until a shadow was cast upon it by the results of certain calculations made by Lord Rayleigh and Sir James Jeans involving the energy radiated by hot objects. The demise of Laplace’s theory was finally witnessed in 1926 when the eminent German scientist Werner Heisenberg advanced the celebrated Uncertainty Principle—the product of the uncertainties in the position and speed of a particle at any time cannot be less than a given amount known as Planck’s constant. Whilst Planck’s constant is small, the Uncertainty Principle has had an inestimable effect on the manner in which physics has evolved and the way the world is viewed. A very small change has thus precipitated far-reaching and significant repercussions— a theme that is central in the study of unstable systems in general, and Chaos Theory in particular. Laplace’s deterministic view of the universe did not just relate to the inanimate. He suggested that the behaviour of individuals could likewise be predicted. Debate about deterministic human behaviour necessarily raises the contentious issue of free will. This nettle will not be grasped here, but suffice it to say that I tend to agree with the comments made by Stephen Hawking at the conclusion of his lecture to the Sigma Club at the University of Cambridge in 1990 [1]. To paraphrase him: ‘everything is determined, but it might as well not be, because we can never know what is determined’. For me then, an individual’s behaviour is the result of his/her free will—but is also subject to the uncertainty and chaos engendered by the influences of nature and other individuals. Whilst examples of the chaos and uncertainty present in everyday life are hardly required, two quite recent personal incidents seem relevant here. At a workshop I attended last September (the Third U.K. Workshop on Software Testing Research), an ‘industrial panel session’ formed part of the second day. The session involved five representatives from industry debating and answering questions from the floor on issues related to the problems of software testing in the real world, and the interaction between academic researchers and industry. One of the representatives made it clear that a central problem in the software industry was that of intense commercial pressure to complete new products, or the next version of existing ones, as soon as possible, and to ‘ship them’ post haste. Subsequently, I felt it necessary to ask the panel the purpose of academics undertaking any research if industry fails to pay attention to or capitalize upon it. In essence, the response was a reinforcement of the problem previously outlined; it left me in no doubt about the intense time and commercial pressures experienced by the software industry, and the almost overwhelming requirement to ‘deliver’. In fact, I began to wonder how long it will be before what appears to be a chaotic situation now is exacerbated to the point that the imperative of swift production and delivery reduces testing to an unjustifiable luxury. For let us not forget that the Second Law of Thermodynamics indicates that entropy is increasing! The second incident concerns a situation that is likely to be familiar to all who have attempted to have their work published. Immediately before attending the workshop referred to above, I learnt of

A Tribute to Martin Woodward

As this issue of STVR was going to press, we were all shocked by the sudden death of Martin Woodward on October 27, 2006. As you can see from the Editorial in this issue written by Lee White, a new Chief Editor had been named on the announcement of the retirement of both Martin Woodward and Lee White as Editors of STVR; that new Chief Editor is Jeff Offutt. Martin was planning to write an Editorial for the next issue, 17(1), of STVR to explain his resignation as Editor, and to share other thoughts on his experiences as Chief Editor of the Journal. Since Martin will not have that opportunity, several of us would like to offer a tribute to Martin and to share some of our experiences with him over the last 13 years.

Editorial: Uncertainty and chaos: Editorials

An acceleration resistant crystal resonator is made from a single quartz plate that is optically twinned into a left-handed (LH) quartz portion and a right-handed (RH) quartz portion. According to the method, the optically twinned single quartz plate is formed into a resonator plate wherein the effective thickness of the (LH) portion is substantially the same as the effective thickness of the (RH) portion. The resonator plate is then mounted and bonded to the support structure of a crystal resonator enclosure. A pair of electrodes is then deposited onto the (LH) portion and a pair of electrodes deposited onto the (RH) portion. The thickness of the two pairs of electrodes is then adjusted so that the resonant frequency of the resonator on the (LH) portion is substantially the same as the resonant frequency of the resonator on the (RH) portion. The pairs of electrodes are then interconnected to each other and to an oscillator circuit so that in-phase vibration of the two portions is assured. The enclosure is then hermetically sealed.

Editorial: Find the missing links?

It was during the summer of 2001, in a meeting with Martin Woodward to discuss plans for STVR, that I agreed to write the editorial for this issue of the journal. Later, when I was thinking about what had transpired at the meeting, it came as quite a shock to realize that, I had not, in fact, undertaken the writing of an editorial for STVR for almost exactly two years! Whilst contemplating this fact, and not without some amazement and regret at so swift a passage of the last two years, I determined to revisit briefly the more recent editorials. Upon so doing, the reason for my two-year ‘sabbatical’ readily became apparent. Over and above editorials addressing such issues as collaboration, specificationbased testing, and changes to STVR and its Editorial Board, each of which was written by one of the ‘regular’ editors, the journal has been fortunate to entertain a number of Guest Editors. In the December 2000 issue, STVR 10(4), John Derrick joined Rob Hierons to host a Special Issue on specification-based testing, while in the June 2001 issue, STVR 11(2), four selected papers from the First International Workshop on Automated Program Analysis, Testing and Verification (WAPATV 2000) were introduced by Nigel Tracey, John Penix and Willem Visser, all members of the workshop’s programme committee. Most recently, however, Eric Wong edited the December 2001 issue, STVR 11(4), introducing three papers from the symposium Mutation 2000 of which he was the programme chairman. Now, apart from enabling me to come to terms with my (obviously distorted) concept of the passage of time, what has all this to do with this issue of STVR? There are in fact several relevant links. First, in STVR 11(2) (see above), Gerard Holzmann and Margaret Smith presented a paper on what currently is, in a number of circles, a ‘hot topic’, namely, model checking. The first article in this issue ‘keeps this pot boiling’ in addressing the issue of efficiency. The paper by Nicola Mazzocca, Antonella Santone, Gigliola Vaglini and Valeria Vittorini is a case study of the use of a methodology for property checking on reduced transition systems in the context of a distributed system (a system for video ondemand). The second paper in this issue, and one that also provides a link to another recent editorial, is a position paper by Rob Hierons, Mark Harman, Chris Fox, Lahcen Ouarbya and Mohammed Daoudi on conditioned slicing. It is noteworthy in a number of ways, including the fact that it is the first position paper to appear in the journal and that it may be viewed as a welcome response to Lee White’s request in his editorial of STVR 9(4) for both papers of this kind and letters to the editor (readers may discern yet another link here!). The third and final article in this issue is a paper by István Forgács and Antonia Bertolino that seeks to address the issue of the untestedness syndrome in respect of data-flow based testing criteria. Specifically, it first defines a new data-flow testing criterion that does not display the syndrome, but may require a path set of infinite size to satisfy it. From this criterion, a second one that

A history lesson

From the Department of Medicine (L.A.B., Z.D.G, S.D.) and the Divisions of Infectious Disease (L.A.B., S.D.) and Cardiology (Z.D.G.), University of Washington School of Medicine, Harborview Medical Center, and the Public Health–Seattle and King County HIV/STD Program (L.A.B.) — all in Seattle; the Department of Internal Medicine, University of Alabama at Birmingham, Huntsville Regional Campus, Huntsville (R.M.C.); and the Veterans Affairs Ann Arbor Healthcare System and Department of Internal Medicine, University of Michigan Medical School, Ann Arbor (S.S.). Address reprint requests to Dr. Dhanireddy at the University of Washington, 325 9th Ave., Box 359930, Seattle, WA 98104, or at sdhanir@uw.edu.