Units of measure as a data attribute

A pragmatic approach to dimensional analysis for mobile robotic programming

“…The dimensional analysis technique can also be used in software, to help make programs more robust against errors in units, by verifying the consistency of the units given. It allows a new class of errors to be caught (Gehani 1977), and improves reliability, writability, and the debugging of errors caused by incorrectly dimensioned data. It can be applied by adding new attributes to data types, for the dimensions and units used, enabling both verification of data compatibility and automatic conversions between different units in the same dimension.…”

“…Numerous dimensional analysis systems have been proposed over many years, including systems for Ada (Gehani 1985;Hilfinger 1988;Rogers 1988;Gonzalez and Peart 1993), Pascal (Gehani 1977;Agrawal and Garg 1984;Dreiheller et al 1986;Baldwin 1987), C++ (Cmelik and Gehani 1988;Umrigar 1994), and Java (Allen et al 2004), proposed regularly since 1977. However, dimensional analysis has still not become a common feature of mainstream languages (although there is a new specification for a future version of Java that specifies a package for dimensioned data classes (Dautelle and Keil 2008)).…”

Rule-Based Analysis of Dimensional Safety

“…The first can appear anywhere in the program and takes as arguments (the underscores) a variable and a unit expression. The variable can be an indexed one in the dynamic version and should be a simple one in the static version; if it is not a simple one then it will be automatically replaced by its simple root, e.g., s [10][i] will be replaced by just s. The unit expression can be any combination of basic units and unit(Expr), the second being evaluated in the current execution environment(s). For example, if x, y, acc, dist and time are variables then the following are admissible assumptions: /*U assume unit(x) = meter U*/ /*U assume unit(x) = unit(y) U*/ /*U assume unit(acc) = meter second^-2 U*/ /*U assume unit(acc) = unit(dist/time) second^-1 U*/…”

“…Karr and Loveman [17] suggested a flexible mechanism that allowed units to occur meaningfully in programs. There have been proposals for dimensional checking within existing languages like Pascal [9,10] and Ada [11], and even in formal specification of software [14]. An intuitive approach to strengthen type checking in programming languages was also suggested in [16].…”

Certifying measurement unit safety policy

“…Mechanisms that allowed units to meaningfully occur in programs were suggested in [14] and [13], and support for measurement unit declarations and checking within existing languages, like Pascal [7,8] and Ada [9], were also proposed. Based on the belief in [21] that type checking can and should be supported by semantic proof and theory, [24] associated numeric types with polymorphic dimension parameters, hereby avoiding measurement unit errors, and a formally verified method to add, infer and check dimension types in ML-style languages was proposed in [16,15].…”