Self-applicable partial evaluation for the pi-calculus

Abstract

In this paper, we are interested in self-applicable partial evaluation for the pi-calculus, a language which models the concurrent behavior of comrnunicat ing processes. We use the claasic three-steps methodology.First, we write a metainterpreter for the language.Second, we introduce an abstract analysis that determines which operations ( communications) can be executed at compile-time.The notion of well-annotatedness of terms is defined. Finally, we exhibit the self-applicable partial evaluator which is applied to wellannotated terms, and we prove its correctness with respect to the interpreter. This approach is compatible with Futamura's projections. Proofs of correctness are baaed on the notion of weak reduction equivalence.

Transformations of CCP programs

“…Notable exceptions are the papers of Ueda and Fukurawa [1988], Sahlin [1995], and of de Francesco and Santone [1996] (their relations with this paper are discussed in Section 7). Also when considering partial evaluation we find only very few recent attempts [Hosoya et al 1996;Marinescu and Goldberg 1997;Gengler and Martel 1997] to apply it in the field of concurrent languages.…”

Partial Evaluation of Concurrent Programs

“…Sequential parts of the programs are written in a functional style. For this kind of programs, partial evaluation allows the static execution of those communications for which the emitter, the receptor and the contents of the message are known [6,9,11]. We obtain a residual program with fewer communications than the original one.…”

“…The authors have proposed a PE for the π-calculus as well as sufficient conditions on the annotations to ensure the correctness of the residual program wrt. the original one [6,12]. Solberg et al and Bodei et al have proposed control flow analyses (CFA) which can be used to improve the precision of the BTA [2,20].…”