Fortran Isolates the CSE Community

The continued usage of Fortran as a dominant programming language isolates the computational science & engineering (CSE) community and hinders the learning and adoption of modern software engineering methods. The issue is not primarily that Fortran is not suitable for developing many different types of software. Indeed, the are many examples of quality CSE software written in Fortran that can legitimately be considered to be “self-sustaining software” (i.e. clean design and code, well tested, minimal controlled dependencies, etc., see [1]). The problem is that there are many CSE development groups and communities that only know Fortran and are therefore isolated from the broader software development community where Fortran usage is extremely low (ranked 24th in the TIOBE Index from Nov. 2013 [11]). Other problems with Fortran will not be discussed in the short paper (e.g. slowness to implement the 2003 standard, lack of an quality open source compiler for Fortran 2003, clumsy and inefficient text processing, etc.). The problem is that if a Fortran programmer wants to learn about quality modern software development techniques and practices (e.g. continuous integration, unit testing, test-driven development (TDD), emergent object-oriented design, refactoring, etc. [2, 9, 8]), to what resources or references do they go? There are few (if any) quality references, training courses or instructors on these topics that are accessible to Fortran-only programmers. The issue is that all of the quality books, articles, training courses on all of these topics have examples in mainstream languages like C++ and Java and not Fortran. Mainstream languages like C++, Java, Python, etc. are so different from Fortran that they are largely inaccessible to Fortran-only CSE programmers and therefore this material is also inaccessible for Fortran-only programmers.