Language interoperability mechanisms for high-performance scientific applications

Language interoperability is a difficult problem facing the developers and users of large numerical software packages. Language choices often hamper the reuse and sharing of numerical libraries, especially in a scientific computing environment that uses a breadth of programming languages, including C, c ++, Java, various Fortran dialects, and scripting languages such as Python. In this paper, we propose a new approach to langauge interoparability for high-performance scientific applications based on Interface Definition Language (IDL) techniques. We investigate the modifications necessary to adopt traditional IDL approaches for use by the scientific community, including IDL extensions for numerical computing and issues involved in mapping IDLs to Fortran 77 and Fortran 90