The object of this investigation is to develop a method for the automatic collection of meaningful statistical information about the causes of program resubmittal in a batch-processing environment. Successive versions of a given program are compared statement-by-statement in order to isolate minor changes made in the program. All statements inserted, substituted, or deleted are examined in terms of (1) the type of statement (i.e. DO, IF, etc.), (2) the number of consecutive statements involved in the change, and (3) how many times the program has already been submitted (the number of “tries”). One thousand and one hundred and ten programs are analyzed in this manner. The method is intended to be used, in conjunction with detailed study of selected cases and with further experimentation in completely controlled situations, to improve programming instruction and manuals, to produce better diagnostic messages, to aid in the design of new compilers, and eventually, to provide means for automatic correction of trivial mistakes. © 1975, Academic Press Inc. (London) Ltd.. All rights reserved.
Mendeley saves you time finding and organizing research
Choose a citation style from the tabs below