Testing software strategy

Testing includes strategies to find software defects and tests designed to prove that the software works. The strategies and tests will provide different results. 

The purpose of testing software is to perform verification and validation. These two concepts are different and understanding that difference is more than just a question of semantics. In order to assure a system and obtain the necessary evidence for the safety case, strategies for both verification and validation need to be in place. 

A CBT/CAT is a test administered at a computer. CBT stands for computer-based test and CAT stands for computer-adaptive test. Special software enables the computer to accurately determine your proficiency at a given task within a short time. In brief, every test taker gets a different set of questions. The computer uses your performance on the early questions to give either easier or more difficult subsequent questions. This is the adaptive part. The harder the questions you successfully answer, the higher your score. You will find more information and specific strategies for CBT/CATs on the following pages. 

Validation plans are documents that tailor a company s overall philosophies, intentions, and strategy to establish performance and computer systems or software adequacy. Validation plans state who is responsible for performing development and validation activities, who identifies which systems are subject to validation, who defines the nature and extent of inspection and testing expected for each system, and who outlines the framework to be followed to accomplish the validation. 

The software test strategy and detailed test scripts (protocol) are reviewed and approved by the Project Team. This review will verify that the test cases will stress the program interfaces, challenge the data boundaries and limits, and verily that the test cases are traceable to the appropriate clauses in the requirements specification deliverable. The review also verifies that the test cases adequately cover the risks identified in the Risk Analysis. The unit/ integration test results are documented by the programmer, and reviewed, approved, signed, and dated by QA. 

Initially, a single test script under the Analyst validation was envisioned. However, as the complexity of the MassChrom software versions was understood, a data migration and system retirement test plan was required to explain the overall strategy with five test scripts. Further information gathering revealed more complexity, and the number of test scripts rose to 10. 

The Validation Plan should also clearly indicate the split of responsibilities between user and supplier. It should be noted that the supplier is normally involved in all levels of testing. The Validation Plan should state that an independent expert (often QA), who will confirm that the testing is being carried out in accordance with the documented test strategy, may witness testing. This is important where the software is developed externally, or where the development is taking place centrally on behalf of several sites. 

This book will acquaint the reader with some of the important tools and strategies now available for investigating chemical processes and systems on personal computers. We present some useful mathematical models, numerical methods for solving them, and statistical methods for testing and discriminating candidate models with experimental data. The software package Athena Visual Studio, obtained from www.AthenaVisual.com, will enable you to do these computations and view the results on your personal computer screen, print them and transmit them. Familiarity with Microsoft Windows (1998 or later) is required, but little knowledge of Fortran, since Athena converts equation statements directly into Fortran code as described in the Athena tutorial on your computer s desktop. 

---