An investigation concerning certain characteristics of Fortran programs



Journal Title

Journal ISSN

Volume Title



It is difficult for language designers and compiler writers to do objective work because it is not known xdiich features of programming languages are most frequently employed by typical users. It is common for a user to frequently employ only a small percentage of the features of a programming language. Since comparatively little information is available about these most frequently used features, the design of languages and compilers must be based on assumptions. It is often true that these assumptions are the most general and not necessarily the most usable. Consequently, statistics regarding the use of programming languages would be of definite value. The primary intent of this research is to provide an example of how statistical data can be used to benefit those working witli programming languages. This thesis concentrates on the techniques that can be used to gather statistical data concerning the usage of the Fortran language, and the benefits that can be realized from the availability of such data. The investigation involves programs from various disciplines, written in EK System/360 Fortran IV G or one of its subsets. A computer program is used to accumulate statistics and to print the results in the form of histograms. Although Fortran is the only programming language investigated, similar techniques are applicable to other languages. The results of this investigation are intended for use in the definition of new programing languages, in writing Fortran compilers, in computer science classes, and by the users of the Fortran language.