The authors have developed a monitor program which is a preprocessor to COBOL programs, which in turn call DBMSs. This monitor is aimed at (1) debugging the user program through data monitoring (tracing), and (2) gathering raw CPU times and IO counts and presenting them as various averages.

The claim to “portability” is based on several facts. First, the monitor itself is written in COBOL and PASCAL, except for the data capture subroutine, which of necessity is in assembly language (ten lines long, not shown). In addition, the whole concept and language used is solidly based upon the CODASYL database proposal.

But the monitor has evidently been implemented only for the UNIVAC 1100, using the DMS1100 DBMS and the DDS1100 data dictionary. The concept of a single monitor which could apply to a variety of DBMSs, COBOLs, and mainframes is a good idea, though one might like to see more experience in moving it around. The statistics generated are limited to various averages of CPU time and IO counts, but the authors suggest a variety of other counts which might be captured in the nonportable assembly language subroutine.

Monitor control statements are added directly to the subject COBOL program via comments (“2 *”/col. 7), and the monitoring printouts they produce seem source-code oriented. Much of this paper is devoted to “internals” of the two-pass precompiler, and there are suggestions that many options have been tried which are not reported upon here. A presumably longer and more detailed paper of a similar title [1] is listed in the Bibliography. All requests for further information should be directed to Stephen R. Schach at the following address: Computer Science Dept., Vanderbilt University, Box 70, Station B, Nashville, TN 37235.

This paper will be of interest to those involved in implementing two-pass preprocessors due to the discussions of “internals” and how problems of implementation were solved. Those interested in monitoring COBOL programs which access DBMSs will find some useful material here, and UNIVAC 1100 users will be especially interested in looking at this well-written and (usually) readable paper.