Catálogo de publicaciones - libros

In this paper we publish the results of a thorough empirical evaluation of a CMM-based software process improvement program that took place at the IT department of a large Dutch financial institution. Data of 410 projects collected over a period of four years are analysed and a productivity improvement of about 20% is found. In addition to these results we explain how the use of linear regression models and hierarchical linear models greatly enhances the sensitivity of analysis of empirical data on software improvement programs.

Deployment of software processes based on reference models is a knowledge-intensive task, i.e., a great amount of technical knowledge must be applied in order to guarantee conformance and adherence of processes deployed to the reference models adopted. Moreover, software process deployers have to deal with organizational and individual cultural problems on a regular basis, for instance, resistances to organizational changes. Therefore, the success of software process deployment within an organization or organizational unit depends on both technical and social aspects of the software process deployment strategy definition and execution. This paper presents the Taba Workstation , an enterprise-oriented Process-centered Software Engineering Environment (PSEE) constituted of an integrated set of tools to support software process deployment based on the Capability Maturity Model Integration (CMMI) and the Reference Model for Brazilian Software Process Improvement (MR-MPS.BR). Software process appraisals demonstrated that the Taba Workstation constitutes one of the most important organizational assets to facilitate the success of software process deployment initiatives and to overcome the inherent difficulties.

The CMMI model for Software Engineering provides guidance for improving an organization’s processes and the ability to develop software systems. The CenPRA test process is a generic software testing model defined by selecting software testing “best practices”; it evolved over the last years and has been published in specific forums. The CenPRA test process, which defines a set of partially ordered activities and test artifacts, has been validated and improved based on the experience of its application at software development companies in Brazil. In this work we carried out an evaluation of the CenPRA test process under the perspective of CMMI. We evaluated essentially which aspects of CMMI are taken into account by the CenPRA test process. We also evaluate how the CenPRA model can be used to supplement software testing related aspects of CMMI. Our results pointed to improvements in the CenPRA test process, and also identify testing tasks and artifacts not considered by CMMI, which can significantly improve an organization testing practices.

Background: Test-driven development (TDD) and pair programming are software development practices popularized by eXtreme Programming methodology. The aim of the practices is to improve software quality. Objective: Provide an empirical evidence of the impact of both practices on package dependencies playing a role of package level design quality indicators. Method: An experiment with a hundred and eighty eight MSc students from Wroclaw University of Technology, who developed finance-accounting system in different ways (CS — classic solo, TS — TDD solo, CP — classic pairs, TP — TDD pairs). Results: It appeared that package level design quality indicators (namely package dependencies in an object-oriented design) were not significantly affected by development method. Limitations: Generalization of the results is limited due to the fact that MSc students participated in the study. Conclusions: Previous research revealed that using test-driven development instead of classic (test-last) testing approach had statistically significant positive impact on some class level software quality indicators (namely CBO and RFC metrics) in case of solo programmers as well as pairs. Combined results suggest that the positive impact of test-driven development on software quality may be limited to class level.

The adoption of agile software development methodologies may appear to be a rather straightforward process yielding instantly improved software in less time and increasingly satisfied customers. This paper will show that such a notion is a misunderstanding and can be harmful to small software development organisations. A more reasonable approach involves a careful risk assessment and framework for introducing agile practices to address specific risks. A case study with a small software development organisation is provided to show the assessment in practice and the resulting risk mitigation strategies for process improvement.

For most XP techniques only a few experimental results on their effects are available. In October 2004 we started a medium-term experiment to investigate the impact of test-first compared to a classical-testing approach. We carefully designed a controlled experiment and conducted it with 18 graduated students randomly assigned to 9 pairs. Hypotheses dealt with development speed, number of test-cases and the test-coverage when applying the testing approaches. Results show differences however not significant ones. This paper also addresses other observations we made during the experimental run. Two major problems strongly affect the results of the experiment: the low number of data points and the non-trivial question, whether students really applied test-first all the time. Although we cannot provide any new results on testing to the research community, this paper contains valuable information about further experimental studies on this topic.

The quality of the software design often has a major impact on the quality of the final product and the effort for development and evolution. A number of quality assurance (QA) approaches for inspection of early-life-cycle documents have been empirically evaluated. An implicit assumption of these studies was: an investment into early defect detection and removal saves higher rework cost. The concept of pair programming combines software construction with implicit QA in the development team. For planning QA activities, an important research question is how effective inspectors can be expected to be at detecting defects in software (design and code) documents compared to programmers who find defects as by-product of their usual construction activities. In this paper we present an initial empirical study that compares the defect detection effectiveness of a best-practice inspection technique with defect detection as by-product of constructive software evolution tasks during pair programming. Surprisingly, in the study context pair programmers were more effective to find defects in design documents than inspectors. However, when building a larger team for defect detection, a mix of inspection and pair programming can be expected to work better than any single technique.

As a key activity in product line engineering (PLE), instantiation is a task to generate target applications by resolving variability embedded in core assets. However, instantiation is often conducted in manual and ad-hoc fashion, largely replying on domain knowledge and experience. Hence, it can easily lead to technical problems in precisely specifying decision model consisting of product-specific variation points and variants, and in handling inter-variant conflicts/dependency. To overcome this difficulty, it is desirable to develop a systematic process which includes a set of systematic activities, detailed instructions, and concrete specification of artifacts. In this paper, we first propose a meta-model of a core asset to specify its key elements. Then, we represent a comprehensive process that defines key instantiation activities, representations of artifacts, and work instructions. With the proposed process, one can instantiate core assets more effectively and systematically.

In the e-Business domain, workflows are central artifacts that are used to specify application systems. To realize reuse at a large scale for e-Business application systems, therefore, workflows need to be reused systematically. To this end workflows must be classified, documented, and stored in a way that enables their identification, evaluation, and adaptation in order to integrate them in an application. Software product line engineering is an established and approved software engineering approach that addresses these issues by handling a number of similar software systems together, enabling large scale reuse during the development and maintenance of the different systems covered by the product line. In this paper, we transfer the concepts of software product line engineering to the domain of e-Business systems by applying the product line techniques to workflows and present initial validation results.

In the mobile industry, system platforms are offered to device developers to enable rapid product development while sharing expensive technology development investments. This paper presents a framework for assessment of requirements engineering collaboration related to statements-of-compliance negotiation in platform subcontracting. The framework includes a classification of platform compliance scenarios and results from analysis of interviews with engineers at two collaborating companies, a device vendor and a platform vendor. Case study findings particular to the compliance scenarios of the framework are provided. The purpose of the framework is to provide a basis for process improvement in collaborative requirements engineering.