Adding Partial Functions to Constraint Logic Programming with Sets
Partial functions are common abstractions in formal specification notations such as Z, B and Alloy. Conversely, executable programming languages usually provide little or no support for them. In this paper we propose to add partial functions as a primitive feature to a Constraint Logic Programming (CLP) language, namely {log}. Although partial functions could be programmed on top of {log}, providing them as first-class citizens adds valuable flexibility and generality to the form of set-theoretic formulas that the language can safely deal with. In particular, the paper shows how the {log} constraint solver is naturally extended in order to accommodate for the new primitive constraints dealing with partial functions. Efficiency of the new version is empirically assessed by running a number of non-trivial set-theoretical goals involving partial functions, obtained from specifications written in Z.