Victor R. Basili – författare

This text presents the knowledge-based approach to the automation of program understanding which plays an important role in nearly all software related tasks: it is vital to the development, maintenance and re-use activities, and is useful in improving the quality of software development. The approach generates program documentation mechanically by combining and building on the strengths of a practical program decomposition method, the axiomatic correctness notation, and the knowledge-based analysis approach. This text is aimed at: computer science students at the advanced graduate level and researchers and practitioners interested in software development, maintenance and re-use.

The importance of properly selecting testing techniques is widely accepted in the software engineering community. However, there are chiefly two reasons why the selections now made by software developers are difficult to evaluate as correct. First, there are several techniques with which the average developer is unfamiliar, often leaving testers with limited knowledge of all the techniques currently available. Second, the available information regarding the different testing techniques is primarily procedure (focused on how to use the technique), rather than pragmatic (focused on the effect and appropriateness of using the technique). The problem addressed in this book is that of improving software testing technique selection.

Engineering tasks are supposed to achieve defined goals under certain project constraints. Example goals of software engineering tasks include achieving a certain functionality together with some level of reliability or performance. Example constraints of software engineering tasks include budget and time limitations or experience limitations of the developers at hand. Planning of an engineering project requires the selection of techniques, methods and tools suited to achieve stated goals under given project constraints. This assumes sufficient knowledge regarding the process-product relationships (or effects) of candidate techniques, methods and tools. Planning of software projects suffers greatly from lack of knowledge regarding the process-product relationships of candidate techniques, methods and tools. Especially in the area of testing a project planner is confronted with an abundance of testing techniques, but very little knowledge regarding their effects under varying project conditions. This book offers a novel approach to addressing this problem: First, based on a comprehensive initial characterization scheme (see chapter 7) an overview of existing testing techniques and their effects under varying conditions is provided to guide the selection of testing approaches. Second, the optimisation of this knowledge base is suggested based on experience from experts, real projects and scientific experiments (chapters 8, 9, and 10). This book is of equal interest to practitioners, researchers and students. Practitioners interested in identifying ways to organize their company-specific knowledge about testing could start with the schema provided in this book, and optimise it further by applying similar strategies as offered in chapters 8 and 9.

Program understanding plays an important role in nearly all software related tasks. It is vital to the development, maintenance and reuse activities. Program understanding is indispensable for improving the quality of software development. Several development activities such as code reviews, debugging and some testing approaches require programmers to read and understand programs. Maintenance activities cannot be performed without a deep and correct understanding of the component to be maintained. Program understanding is vital to the reuse of code components because they cannot be utilized without a clear understanding of what they do. If a candidate reusable component needs to be modified, an understanding how it is designed is also required. of This monograph presents a· knowledge-based approach to the automation of program understanding. This approach generates rigorous program documentation mechanically by combining and building on strengths of a practical program decomposition method, the axiomatic correctness notation, and the knowledge­ based analysis approaches. More specifically, this approach documents programs by generating first order predicate logic annotations of their loops. In this approach, loops are classified according to their complexity levels. Based on this taxonomy, variations on the basic analysis approach that best fit each of the different classes are described. In general, mechanical annotation of loops is performed by first decomposing them using data flow analysis. This decomposition encapsulates interdependent statements in events, which can be analyzed individually.

Engineering tasks are supposed to achieve defined goals under certain project constraints. Example goals of software engineering tasks include achieving a certain functionality together with some level of reliability or performance. Example constraints of software engineering tasks include budget and time limitations or experience limitations of the developers at hand. Planning of an engineering project requires the selection of techniques, methods and tools suited to achieve stated goals under given project constraints. This assumes sufficient knowledge regarding the process-product relationships (or effects) of candidate techniques, methods and tools. Planning of software projects suffers greatly from lack of knowledge regarding the process-product relationships of candidate techniques, methods and tools. Especially in the area of testing a project planner is confronted with an abundance of testing techniques, but very little knowledge regarding their effects under varying project conditions. This book offers a novel approach to addressing this problem: First, based on a comprehensive initial characterization scheme (see chapter 7) an overview of existing testing techniques and their effects under varying conditions is provided to guide the selection of testing approaches. Second, the optimisation of this knowledge base is suggested based on experience from experts, real projects and scientific experiments (chapters 8, 9, and 10). This book is of equal interest to practitioners, researchers and students. Practitioners interested in identifying ways to organize their company-specific knowledge about testing could start with the schema provided in this book, and optimise it further by applying similar strategies as offered in chapters 8 and 9.

Program understanding plays an important role in nearly all software related tasks. It is vital to the development, maintenance and reuse activities. Program understanding is indispensable for improving the quality of software development. Several development activities such as code reviews, debugging and some testing approaches require programmers to read and understand programs. Maintenance activities cannot be performed without a deep and correct understanding of the component to be maintained. Program understanding is vital to the reuse of code components because they cannot be utilized without a clear understanding of what they do. If a candidate reusable component needs to be modified, an understanding how it is designed is also required. of This monograph presents a· knowledge-based approach to the automation of program understanding. This approach generates rigorous program documentation mechanically by combining and building on strengths of a practical program decomposition method, the axiomatic correctness notation, and the knowledge­ based analysis approaches. More specifically, this approach documents programs by generating first order predicate logic annotations of their loops. In this approach, loops are classified according to their complexity levels. Based on this taxonomy, variations on the basic analysis approach that best fit each of the different classes are described. In general, mechanical annotation of loops is performed by first decomposing them using data flow analysis. This decomposition encapsulates interdependent statements in events, which can be analyzed individually.