»System testing should be automated as well. A collection of executable programs should be produced and maintained to exercise all parts of the system. The set should be open ended and maintenance utilities should be included. A result (i.e. output) from a system source text update should be a set of necessary control statements to enable a selective test of all modules modified and referenced entities. A control sequencer utility should exist which would selectively execute all programs to exercise specified modules. Test codes should exercise modules separately and in combination. A total system stability and performance test should be included in such a scheme. Driving the system from this source should be recoverable in event of an error so that a thorough test can be made in one pass through the test collection. Errors should be diagnosed and reported, tracing the chain of events which resulted in the error, and should attempt to isolate the causal factor to a module or referenced element. The test set should be maintained in source text language by the same facility discussed with reference to system source text. Clearly there is a need to be able to include the system module and change directories as a part of the test source environment and to be able to cross reference to these entries from within non-generative code in the source text of the test programs. Tests should be dated to permit exclusion of their usage when they become obsolete. As an output of a test validation run, each test should list the modules it has exercised, and as well, should list the interfaces and tables it has tested. It is important to document success, as well as failure.«