Call pattern analysis for functional logic programs

Abstract

This paper presents a new program analysis framework to approximate call patterns and their results in functional logic computations. We consider programs containing non-strict, nondeterministic operations in order to make the analysis applicable to modern functional logic languages like Curry or TOY. For this purpose, we present a new fixpoint characterization of functional logic computations w.r.t. a set of initial calls. We show how programs can be analyzed by approximating this fixpoint. The results of such an approximation have various applications, e.g., program optimization as well as verifying safety properties of programs.

A compact fixpoint semantics for term rewriting systems

“…We consider a set of benchmark programs that satisfy the conditions for the correctness and completeness of our approach. The benchmarks used for the comparison are: risers and tails, which are two almost-orthogonal and terminating Haskell programs that are commonly used for analyzing the safety of pattern matching in Haskell [34]; doubleisone, an almost-orthogonal and terminating TRS for doubling and unity checking borrowed from [24]; bertconc, a right-linear, canonical TRS for concatenating lists borrowed from [7]; vending, a topmost TRS adapted from [10]; and insertsort, a standard encoding of insertion sort borrowed from the Termination Problem DataBase. 12 The source code of these benchmark programs is available at the prototype web site.…”

Functional logic programming

“…The formal foundations of functional logic languages are also useful for the development of environments and tools for reasoning about programs. Developments in these areas include tools for program analysis [23], verification [16], partial evaluation [1], profiling [11], debugging [10], and testing [17].…”