It is widely accepted that early reviews on requirements specifications (RS) are an effective and efficient quality assurance technique. So why are they still not applied all over the software industry? In this paper we pinpoint that this is due to five major challenges: 1) Software requirements are based on flawed `upstream' requirements and reviews on RS are thus in vain. 2) The impact of sociological issues related to reviews is underestimated. 3) Important quality aspects of RS escape reviews. 4) The goal of applying reviews is not made clear and different review approaches are mixed. 5) Incremental software development poses specific challenges to applying reviews on RS. In this paper we argue that in order to solve these five challenges research on reviews must take a more holistic approach, stretching to pre-project phases and incorporating various other disciplines in order to add more value for the software industry. The paper also offers preliminary solutions to the discussed challenges and sketches open research questions of high relevance for the software industry.

To be successful with global software development (GSD), development knowledge needs to be shared among the developers and stakeholders, and the quality of the exchanged information must be assured. Therefore, mature processes, methods and tools have to be in place. If a unified and integrated solution does not exist, this impedes the exchange of knowledge (and the migration of people between projects). In GSD, such a diversity can lead to new problems: offshore development teams have to repeatedly re-adjust to method variants used by the respective business units. This can lead to misinterpretation of information and risks for project success. We report on re-aligning the varying software engineering methods and unifying the methodology throughout Capgemini sd&m. We also standardized quality assurance procedures and tightly integrated them with the engineering methodology. By this, we arrived at a comprehensive company-wide Enterprise Software Engineering Model that effectively supports knowledge transfer from clients to the onshore and offshore team.

Effective knowledge transfer (KT) is always important in software development projects, but crucial in global software development (GSD). Two challenges arise: First, reviews of the software requirements specification (SRS) are indispensable, but not always effective. Second, using knowledge representations that support KT from customers to developers is paramount. However, 'classical' SRS often don't support SRS comprehension of all stakeholders. We address these two challenges with a new approach that exploits the multi-fold power of a acceptance test case specifications (ATC-Specs): 1) A specific two-stage test-based review technique is used. We argue that these two-stage reviews of ATC-Specs increase the quality of the ATC-Specs and the SRS. 2) Additionally to the SRS, ATC-Specs are delivered to the offshore team, bridging the mental models of different stakeholders, and thus effectively transferring knowledge. We provide preliminary evidence of the validity of our approach based on a commercial GSD project at Capgemini sd&m.

Assessments are an effective technique for software quality assurance. As global software development (GSD) becomes the standard, an assessment framework must be flexible to support different sourcing and shoring models. Although much work exists on inspections and reviews, an assessment framework which addresses these challenges is missing. We present a systematic yet flexible assessment framework. The paper contributes: i) The description of our assessment framework which addresses four challenges: Appropriateness of a software requirements specification (SRS), viability of software architectures and SRS, wholeness of work packages, and compliance of results with predefined quality objectives. ii) A detailed explanation how the assessment framework can be tailored to support offshore and outsourcing scenarios. This paper describes the result of a two years research initiative at Capgemini sd&m and serves the practitioner to implement assessment frameworks according to his needs. We also discuss open research questions of high relevance for the software industry.