Fortran Programming Language and Scientific Programming: 50 Years of Mutual Growth

Abstract

This special issue of Scientific Programming is devoted to celebration of fifty years of mutual and remarkable growth of both scientific programming and its primary language, Fortran. In this brief introduction, we will remark on the past, the present and the future of Fortran and summarize the papers included in this issue. Although the first specification of the Fortran language was released in 1956, IBM delivered its first compiler for its computer, IBM model 704, in 1957, hence this year marks the 50 anniversary of introduction of Fortran to users. The language was designed by John Backus and his colleagues at IBM with the goal to reduce the cost of programming scientific applications by providing an “automatic programming system” to replace assembly language with a notation closer to the scientific programming domain. As the computer technology has been evolving from a single computer, to parallel computers of different kinds, to multi-core processors, clusters and grids, the Fortran language has been evolving as well with the same goal of reducing the cost of programming without sacrificing efficiency. Over the years, the common standard emerged first, followed by a series of revisions. This process continues as the most recent development of a revision, scheduled for 2008, is under way. Yet, care has been taken at each revision to preserve, to the greatest extent possible, compatibility with previous versions to ensure correct recompilation of legacy codes. As a result, each revision includes only a fraction of the proposed new facilities, allowing the ideas for changes to mature before they are accepted. Over half of the century of its existence, the evolving Fortran has been the traditional and major language for scientific programming and it has played a significant role in the research on programming languages and compilers for scientific computing. The vibrant user community, the wellestablished committee overseeing its evolution and the process for revisions of the language ensure the significant role for Fortran in scientific programming in the immediate future and are likely to continue to secure Fortran’s relevance for many years to come. The first article in this issue, entitled “Scientific programming in Fortran” has been prepared by W. Van Snyder. The author provides a historical perspective on the development of various, progressively more modern, versions of Fortran, starting with the first standard developed by ANSI predecessor, ASA, and introduced in 1966. The article briefly describes the essence of revisions to these standards introduced in 1977, 1990, 1995 and most recently in 2003. Then, the author briefly discusses revisions being prepared for the 2008 standard release and remarks: “Of the thousands of programming languages invented, only a dozen or so have been sufficiently widely used to have had international standards, and only four of those have had revisions that attempted to keep up with language technology while preserving software investments by maintaining compatibility with previous editions: Fortran, Cobol, Ada and C . . . The temperament of the Fortran committees and community, at least at present, is that there will be future developments, which will maintain, and probably enhance, the suitability of Fortran for scientific programming.” This clearly shows how unusual the longevity and vitality of Fortran have been.