School Homepage

Papers of Patricia M. Hill

Software Support for CLP: Papers

Technical Reports at Leeds

Efficient Structural Information Analysis for Real CLP Languages

[Page last updated on 2000/10/25.]

Roberto Bagnara
Dipartimento di Matematica
Università di Parma
Via D'Azeglio, 85/A
I-43100 Parma
Italy

Patricia M. Hill
School of Computing
The University of Leeds
Leeds LS2 9JT
England

Enea Zaffanella
Dipartimento di Matematica
Università di Parma
Via D'Azeglio, 85/A
I-43100 Parma
Italy

Abstract:

We present the rational construction of a generic domain for structural information analysis of real CLP languages called Pattern(D#), where the parameter D# is an abstract domain satisfying certain properties. Our domain builds on the parameterized domain for the analysis of logic programs Pat(R), which is due to Cortesi et al. However, the formalization of our CLP abstract domain is independent from specific implementation techniques: Pat(R) (suitably extended in order to deal with CLP systems omitting the occurs-check) is one of the possible implementations. Reasoning at a higher level of abstraction we are able to appeal to familiar notions of unification theory. This higher level of abstraction also gives considerable more latitude for the implementer. Indeed, as demonstrated by the results summarized here, an analyzer that incorporates structural information analysis based on our approach can be highly competitive both from the precision and, contrary to popular belief, from the efficiency point of view.


Available: Gzipped Postscript, BibTeX entry.