Re-engineering legacy mission scientific software
Charles D. Norton, Viktor K. Decyk · AIAA Space 2001 Conference and Exposition · 2001
Many mission-critical scientific applications often rely on a legacy of software representing great intellectual and commercial value. This software is generally well debugged, produces trusted results, is actively meeting end-user goals, and preserves (sometimes hidden) expert knowledge that cannot be easily reproduced. Nevertheless, more ambitious missions require increased capabilities that impose new demands on software. Should these legacy codes be abandoned and rewritten from scratch, or can they be modernized to achieve new objectives? Our approach for modernizing legacy scientific software, based on the new features of Fortran 90/95, will be presented. The methodology adds new capabilities, and increased safety, while promoting collaborative development and abstractionbased design. Application of this technique to modernize the Modeling and Analysis for Controlled Optical Systems software (developed at JPL and important to NASA's Next Generation Space Telescope Project) will be described concluding with new directions such as software tools for partial automation and evolution toward object-oriented concepts. MODERNIZING SCIENTIFIC SOFTWARE