Abstract The widespread acceptance of parallel and distributed processing will result in migration of millions of lines of sequential code to parallel environments over the next decade. Unlike movement across sequential environments, significant structure and code changes must be made in order to benefit from the increased processing power of parallel computing. Nowhere is this problem more obvious than in the scientific computing community. The sequential nature of FORTRAN has limited its ability to utilize the performance achievable from parallel and/or distributed systems. Strand, a concurrent logic programming language based on Edinburgh Prolog, is intended to exploit concurrency in both parallel and distributed environments, while offering portability across a variety of platforms and topologies. Strand minimizes reprogramming of existing software by providing a Foreign Language Interface (FLI) which allows inclusion of C and/or FORTRAN code in Strand programs. This paper describes a set of tools called the pRETS (parallel Reverse engineering ToolSet) which supports the semi‐automated conversion of FORTRAN programs into Strand Foreign Language Kernels.

The authors discuss a parallel reverse-engineering tool set (pRETS) to support the adaptation of sequential Fortran programs to parallel processors. The tool set is based on a Prolog database representation of the program. The database contains dataflow dependency information for the application program, as well as its complete abstract syntax tree. The database can be queried interactively to extract design information about the program that is being adopted. Queries about the program are formulated with Prolog clauses. A set of standard queries for common information requirements has been constructed.>