Mercury is a logic programming language that is considerably faster than traditional Prolog implementations, but lacks support for full unification. HAL is a new constraint logic programming language specifically designed to support the construction of and experimentation with constraint solvers, and which compiles to Mercury. In this paper we describe the HAL Herbrand constraint solver and show how by using PARMA bindings, rather than the standard WAM representation, we can implement a solver that is compatible with Mercury’s term representation. This allows HAL to make use of Mercury’s more efficient procedures for handling ground terms, and thus achieve Mercury-like efficiency while supporting full unification. An important feature of HAL is its support for user-extensible dynamic scheduling since this facilitates the creation of propagation-based constraint solvers. We have therefore designed the HAL Herbrand constraint solver to support dynamic scheduling. We provide experiments to illustrate the efficiency of the resulting system, and systematically compare the effect of different declarations such as type, mode and determinism on the resulting code.
|Title of host publication||Program Development in Computational Logic|
|Subtitle of host publication||A Decade of Research Advances in Logic-Based Program Development|
|Editors||Maurice Bruynooghe, Kung-Kiu Lau|
|Place of Publication||Berlin Germany|
|Number of pages||40|
|Publication status||Published - 2004|
|Name||Lecture Notes in Computer Science|