Good test-suites are an important tool to check the correctness of programs. They are also essential in unsupervised educational settings, like automatic grading or for students to check their solution to some programming task by themselves. For most Haskell programming tasks, one can easily provide high-quality test-suites using standard tools like QuickCheck. Unfortunately, this is no longer the case once we leave the purely functional world and enter the lands of console I/O. Nonetheless, understanding console I/O is an important part of learning Haskell, and we would like to provide students the same support as with other subject matters. The difficulty in testing console I/O programs arises from the standard tools’ lack of support for specifying intended console interactions as simple declarative properties. These interactions are however essential in order to determine whether a program behaves as desired. We describe the console interactions of a program by tracing its text input and output actions. In order to describe which traces match the intended behavior of the program under test, we present a formal specification language. The language is designed to capture interactive behavior found in commonly used textbook exercises and examples, or as much of it as possible, as well as in our own teaching, while at the same time retaining simplicity and clarity of specifications. We intentionally restrict the language, ensuring that expressed behavior is truly interactive and not simply a pure string-builder function in disguise. Based on this specification language, we build a testing framework that allows testing against specifications in an automated way. A central feature of the testing procedure is the use of a constraint solver in order to find meaningful input sequences for the program under test.

Body fat mass (FM) adds to the variance in resting energy expenditure (REE). However, the nature and extent of this relationship remains unclear. Using a database of 1306 women and a linear regression model, we systematically analysed the contribution of FM to the total variance in REE at different grades of adiposity (ranges of body %FM). After adjusting for age, the relative contribution of FM on REE variance increased from low ( ≤ 10 %FM) to normal (>10– ≤ 30 %FM) and moderately elevated (>30– ≤ 40 %FM) grades of adiposity but decreased sharply at high (>40– ≤ 50 %FM) and very high (>50 %FM) grades of adiposity according to the ratio between regression coefficients. These data suggest that the specific metabolic rate of fat tissue is reduced at high adiposity. This should be considered when REE is normalized for FM in obesity.

Current anthropometric indices for health risk assessment are indirect measures of total or visceral body fat mass that do not consider the inverse relationship of lean body mass to metabolic risk as well as the non-linear relationship between central obesity and insulin resistance.We examined a new anthropometric index that reflects the relationship of waist circumference (WC) as a risk factor to fat-free mass (FFM) as a protective parameter of body composition. In apopulation of 335 adults (191 females and 144 males; mean age 53 (sd 13·9) years) with ahigh prevalence of obesity (27%) and metabolic syndrome (30%) we derived FFM:WC3 from the best fit of the relationship with metabolic risk factors (plasma triacylglycerol levels and insulin resistance by homeostasis model assessment index). Because FFM is known to be proportional to the cube of height, FFM was subsequently replaced by height3 yielding height3:WC3 as an easily applicable anthropometric index. Significant inverse relationships of height3:WC3 to metabolic risk factorswere observed for both sexes. They slightly exceeded those of conventional anthropometric indicessuch as BMI, WC or WC:hip ratio in women but not in men. The exponential character of the denominator WC3 implies that at a given FFM with gradually increasing WC the increasein metabolic risk is lower than proportional. Further studies are needed to evaluate height3:WC3 as an anthropometric index for health risk assessment.