Automated testing in APL—an application of exception handling

Abstract

Several APL systems have recently been enhanced to include exception handling facilities. The construction of an automated tester is presented as a case study of one such definition: the latent expression exception handling system [1] available on STSC's APL*PLUS@@@@ System. The automated tester was developed as a tool to manage the systematic verification of APL system features through programmed tests. A test is an APL program containing assertions that prescribe the outcome of associated statements. The tester verifies that the asserted outcomes are in fact realized during execution. This paper describes the central role of exception handling in the implementation of the automated tester. When tests are executed as APL programs, the exceptions generated by the assertion construct can be used to control interpretation of the assertion. Other errors encountered during test execution must be intercepted and recorded, so that it is possible to assert that a statement results in an error. Finally, since the tester should execute large sets of tests unattended, it should not be interrupted by improper tests or unexpected results.