IEEE Software

PrePrint: Handshaking: Negotiate to Provoke the Right Understanding of Requirements

Development teams that accept fragmentary and vague customer requirements have to cope with uncertainty and disorientation. Inadequate communication and submission to a dominating customer makes it hard to know if requirements are understood or not. Requirements engineering, with its focus on good specification practices, had yet to find working solutions for effective requirements communication. We are using a negotiation process called Handshaking with Implementation Proposals to make such requirements communication work – even in situations where written requirements are almost nonexistent and distance separates customers from developers. Handshaking is an efficient and effective technique that uses architectural options to achieve requirements understanding, to create value with implementation decisions, and to establish the foundation for a stable project. This article focuses on the interface between product management and the development organization in companies, where product management acts as a customer towards development. We describe the requirements communication challenges we encountered, explain how we addressed these challenges with the Handshaking process, and share the lessons we learned at European and Asian development sites from ABB and Danaher Motion.

PrePrint: Software industry performance – ‘what you measure is what you get’

Software industry overall performance is uneven and at first sight puzzling. Delivery to time and budget is notoriously bad and productivity shows limited improvement over time, yet quality can be amazingly good. Customers largely bear the costs of the poor aspects of performance. Many factors drive this performance. This paper explores connections which suggest there may be causal links between (a) the commonly-used performance metrics, estimating methods and processes, (b) the incentives placed on suppliers, and (c) the resulting observed performance. The paper proposes a set of possible improvements to current metrics and estimating methods and processes, and concludes that software professionals must educate their customers on the levers that are available to obtain a better all-round performance from their suppliers.

PrePrint: Tool Support for Change-centric Test Development

Developers use unit testing to improve the quality of software systems. Current development tools for unit testing help to automate test execution, to report results, and to generate test stubs. However, they offer no aid for designing tests aimed specifically at exercising the effects of changes to a program. This paper describes a method for applying change impact analysis to test-driven development, to provide developers with quantitative feedback of test coverage of their changes. This information can be used to meet defined coverage goals or to help generate new tests to reveal unanticipated change effects. The approach, called "change-centric test development", is tool supported; a typical scenario shows the effectiveness of our tool JUnitMX in a practical feasibility study.

PrePrint: Experiences of Using the Agile Unified Process in the Banking Sector

Agile Unified Process (AUP) is a recently introduced public domain customization of Rational Unified Process. In this paper we present an experience report of the application of AUP in a software development project in the banking sector exploiting Service-Oriented Architecture (SOA) functionality and user interface integration of client server applications. We describe the project in detail and present the conditions and steps required so that Agile Methods can be successfully combined with RUP process framework resulting in a light-weight and flexible approach to development. Using AUP, the project was successfully delivered within all budgetary and time-constraints set forth. We conclude with a series of lessons learnt from the successful application of the Agile Unified Process in SOA-based projects in banking.

PrePrint: Operational and Strategic Learning in Global Software Development - Implications from two Offshoring Case Studies in Small Enterprises

Small to medium-sized software enterprises (SME) increasingly participate in offshoring activities. Detecting market niches and deploying highly flexible software development approaches are seen as key competitive abilities of SME. Therefore, it is of major importance to learn how offshoring affects these capabilities which are closely related to organizational learning. We present case studies from two German companies that engage in offshoring of software development. By comparing the cases with each other, we highlight the different structures the companies chose for their development work and how these structures were enacted in practice. Furthermore, we show how related practices affect strategic and operational aspects of Argyris et al.’s (1985) conception of single- and double-loop learning. Our case studies show that organizational learning is a problem for SME engaged in offshoring and that an inability for double-loop learning can even lead to failures in case of organizational restructuring.

PrePrint: Architecture As Language

This article showcases and explains the use of domain-specific languages to express software architectures. Specifically, I report about project experiences where we created system-specific textual DSLs that were used to describe the system in a tool-processable way, to reason about the system, and to generate the majority of technical implementation code. The article starts off with a quick look at software architecture today and introduces the concept of architecture DSLs. Then follows an extensive example that shows an example language and how it evolved together with the architecture it describes. The next major section looks at the benefits of the approach, discusses why we used textual languages as opposed to graphical ones, and discusses issues around tooling, model validation, code generation, and the importance of standard modeling languages. The article concludes by taking a brief look at some of the challenges of the approach.

PrePrint: Theory of Relative Dependency: Higher Coupling Concentration in Smaller Modules and its Implications for Software Refactoring and Quality

Our recent studies have repeatedly found that smaller modules are proportionally more defect-prone. In this paper, we formulated and tested a hypothesis stating that smaller modules are proportionally more coupled given that dependencies caused by coupling has been consistently associated with defect-proneness. We found strong evidence supporting this hypothesis. Further, we found that refactoring practices exacerbate this effect. Based on the highly consistent results obtained in this study, we state an empirically-based theory for software modules called the theory of relative dependency: In large scale software systems, smaller modules will be proportionally more dependent compared to larger ones. The implications of our findings for practice are twofold: (1) we now have an empirically supported mechanism explaining the observations that defect concentration is higher in smaller modules, which can be used by practitioners as an evidence while seeking resources and support to revise or amend the existing quality assurance and quality control practices in their organizations; (2) for projects that refactor extensively, such as those using agile methods, focusing defect detection and correction activities on smaller modules will lead to even more effective defect detection.

PrePrint: Supporting Trust in Distributed Teams through Continuous Coordination

Abstract. In this paper we report our investigation of the intersection of two areas: trust in distributed development teams and the role that software tools can play in the support of teams. We report the findings of a field study we performed as well as the conclusions we derived from an extensive review of the literature concerning trust in teams. These findings are collectively used as a means to investigate the role that software tools play in promoting trust. Specifically we review tools that were developed based on the Continuous Coordination (CC) paradigm that we developed in support of distributed development. Our investigation led us to conclude that these tools (Palantír, Ariadne, World View, and Workspace Activity Viewer) do play a role, primarily by sharing information across boundaries, through visualizations, and in other ways relevant to tool characteristics and related to their visualizations. Our analysis provides insights into the role existing tools play in developing trust and can also provide insights on how future tools can be designed to promote trust in distributed teams.

IEEE Software - November/December 2009 (Vol. 26, No. 6)

IEEE Software

PrePrint: Tools for Continuously Evaluating Distributed System Qualities

The end-to-end evaluation of quality-of-service (QoS) properties (e.g., performance, reliability, and security) for distributed systems has historically occurred late in the software lifecycle. As a result, many design flaws that affect QoS are not found and fixed in a timely and cost effective manner. This article shows how model-driven engineering—particularly domain-specific modeling languages coupled with system execution modeling tools—can enable agile development of distributed systems and facilitate continuous system integration testing to improve quality assurance of QoS properties throughout the software lifecycle.

PrePrint: Star Search – a Light-weight Innovation Process for Software Intensive Product Development

The product development environment facing most companies today requires having a long-term perspective featuring the conception and development of long-term innovations. This can be hard when close quarter bottom-line results dominate. The problem is that without innovation the competitive advantage will decrease over time. This is especially true for companies producing software intensive systems. Software is becoming a large part of the competitive advantage of traditionally hardware-focused systems such as cars, robots, or power systems where feature sets traditionally offered and controlled by hardware are transferred to software. As the impact and influence of software grows, so does the possibilities for innovation and increasing the competitive advantage through software. This paper presents Star Search, a lightweight innovation model developed based on best practices from innovation management literature as well as two industry cases. Star Search utilizes face-to-face screening and refinement of ideas using homogenous audition teams. Star Search was developed in collaboration with, and subsequently piloted at the two companies. The result show promising indications that Star Search has helped increase the long-term perspective of product development by increasing the level of new ideas that make it to actual product planning and development.

PrePrint: Recommendation Systems for Software Engineering

Software development can be challenging because of the large information spaces that developers must navigate. Without assistance, developers can become bogged down, and spend a disproportionate amount of their time seeking information at the expense of other value-producing tasks. Recommendation Systems for Software Engineering are software tools that can assist developers with a wide range of activities, from reusing code to writing effective bug reports. We provide an overview of recommendation systems for software engineering: what they are, what they can do for developers, and what they might do in the future.

PrePrint: Using a Line of Code Metric to Understand Software Rework

Rework should be factored into sizing and/or productivity calculations when creating software effort estimates. This paper defines reworked code as the size of software that is created during development but does not exist in the final build. The utility of using lines of code as a sizing metric was explored and found to be relevant under criteria such as similar projects, process, development team, and coding constraints. A simple method was implemented that frequently measured new effective lines of code (nLOC) – added, changed or deleted non-blank non-comment lines of code - during software development. The quantity of nLOC was measured bi-weekly and accumulated over the project’s development. It was also calculated between initial and final builds and then subtracted from accumulated nLOC to determine reworked code. Three projects were analyzed and indicated that between 19% and 40% of the code written was not in the final release.

PrePrint: A Process for Managing Risks in Distributed Teams

Distributed software projects represent particular risks that need careful managerial attention. We present a process for managing these risks. The process is based on state-of-the-art literature on distributed team risks, resolution techniques addressing these risks, and guidelines for how resolution techniques apply to risks. Following CMMI’s generic approach to risk management, the process offers a series of rigorous steps that are readily understood and easy to follow. We illustrate the process based on a large-scale, strategic software project with multiple subprojects that cross organizational and national boundaries. In conclusion, we provide suggestions for tailoring the process to a particular project or organization.

PrePrint: Domain Specific Languages in a Customs Information System

This paper presents a case study in applying the domain-specific languages (DSL) approach for building a configurable customs information system, namely the Customs Engine. The Customs Engine consists of several subsystems built on top of a componentized platform. Each component of the platform is divided into two layers: a formal specification of the component written in a DSL specific to that component, and an implementation of the DSL in question. The use of DSLs allowed us to follow a highly iterative and reuse-oriented development method. Our analysts benefited from the ability to specify the behaviour of the system directly, instead of relaying this information to programmers using lengthy human-language documentation. However, we encountered problems when trying to enable non-technical domain experts to write or modify DSL programs by themselves.

PrePrint: Theory of OO Analysis - Is it just Theory ?

Academic researchers and practitioners alike frequently describe a situation in which the Object Oriented (OO) approach strongly supports the design phase but only weakly supports the functional analysis phase. Such observations conflict with impressions imparted in textbooks and training programs. The current study examines whether Object Oriented Analysis (OOA) theory has been deployed in practice (with or without the support of CASE tools) or whether it remains theoretical. This study explores 54 different projects that range widely in scope and topic, all of which were implemented using OO tools. It analyzes the situation in the field, indicates potential pitfalls and suggests directions for cost–benefit improvement. We show that the OOA has remained strong in theory but has not been widely adopted by practitioners. We suggest cost-benefit interpretation, which presume that OOA fails to attract practitioners because of low ratio of cost-benefit relative to other methodologies. In particular, the cost of system modeling using OOA methodology and all UML diagrams is perceived to be high relative to its benefits. A number of recommendations for cost-effectiveness are presented.

PrePrint: Mining for Computing Jobs

Understanding the types of jobs, the skills utilized in those jobs, and the relative distribution of jobs currently available within the Computing professions is imperative in today’s economy. This research project uses a web content mining approach to address these information needs. With this approach, the authors have identified twenty categories of jobs and their associated skills needs which are currently prevalent in the Computing professions. During the study period, nearly a quarter million job ads were extracted from monster.com, hotjobs.com, and simplyhired.com and subsequently analyzed. This paper describes the process in which the jobs were extracted and analyzed as well as a discussion of the different types of jobs which were identified. The resulting data will be useful to current computing professionals, prospective employees, human resource executives, and educational institutions.

PrePrint: Harnessing Web 2.0 for ERP Implementations

ERP implementations cost billions of dollars annually in United States alone. This article describes how Web 2.0 may be used to reduce the cost, improve the quality and lower the risk of ERP implementations. Challenges and requirements for a web 2.0 knowledge repository system are gathered from ERP implementation communities through a case study. To fulfill these requirements, the article presents EPICS, a web 2.0 system that supports knowledge reuse in ERP implementations. A prototype is built upon a number of popular open source software packages. Prototype deployment at a real-world ERP community shows promising results. Besides cost savings, interesting business models emerge as a result of the deployment.

PrePrint: Static Architecture Conformance Checking -- An Illustrative Overview

In this paper, we compare and illustrate the use of three static architecture conformance techniques, namely dependency structure matrices, source code query languages, and reflexion models. To highlight the similarities and differences between the three techniques, we describe how some of their available supporting tools can be applied to specify and check architectural constraints for a simple personal information management system.

PrePrint: Operational and Strategic Learning in Global Software Development - Implications from two Offshoring Case Studies in Small Enterprises

Small to medium-sized software enterprises (SME) increasingly participate in offshoring activities. Detecting market niches and deploying highly flexible software development approaches are seen as key competitive abilities of SME. Therefore, it is of major importance to learn how offshoring affects these capabilities which are closely related to organizational learning. We present case studies from two German companies that engage in offshoring of software development. By comparing the cases with each other, we highlight the different structures the companies chose for their development work and how these structures were enacted in practice. Furthermore, we show how related practices affect strategic and operational aspects of Argyris et al.'s (1985) conception of single- and double-loop learning. Our case studies show that organizational learning is a problem for SME engaged in offshoring and that an inability for double-loop learning can even lead to failures in case of organizational restructuring.