Dukas: An environment for software process description and guided enactment

The focus of the research reported in this thesis comes about from the synthesis of two existing fields relating to Software Engineering. The first field is that of Process-centred Software Engineering Environments (PSEE) for which the goal is to develop environments whose major concern is for the processes by which software systems are developed. Therefore, in addition to software engineering activities, a PSEE should support process engineering - the analysis, design, implementation and evaluation of software process models. The second field is that of Software Process Modelling. Software Process Modelling (SPM) can be defined as the definition of models of a software development process that may be used for a number of purposes, for example to understand, enact, analyse and improve them. Thus, since a PSEE requires processes to be modelled in some way for it to execute, and enacting a software process model, for the purposes of analysis and improvement, requires an enaction environment, we have two complementary areas of research. We propose this is a novel research area that has not been previously addressed in the literature; the PSEE literature assumes that models exist in a form relevant to their needs, and the SPM literature contains no discussion of existing environments for capturing information about models in order to understand them. Our research, as we report in this thesis, explores research issues from each of the two fields and uses the results to achieve two, quite specific, goals, one of which relates to SPM and the other to PSEE. Our first goal was 'objective-driven' Software Process Modelling, the second was a Software Process Task Management Environment (SPTME). For 'objective-driven' SPM, the goal was to develop one of more methods with which to model desirable software processes that can be initially derived from an existing process reference model, and then adapted. The reference model provides an initial set of parameters and activities that are used to create a concrete model of the process for use by a project-based, development team. Our second goal was to draw on PSEEs concepts to develop a set of requirements for a support environment that could handle the day-to-day management of the processes as defined in our concrete models in order to provide any information necessary to keeping a project on track, and for continuous analysis and improvement of the process models. We postulate that a PSEE that is designed and developed with an understanding of the objectives of the software process models that it will contain, and potentially enact, is likely to be found more acceptable to its users than one that is not. Similarly, objective-based software process modelling is more likely to be adopted, in a practical sense, if there are support environments that are sympathetic to its requirements. Our approach to achieving our research goals appears to differ from other reported work in the field of process-centred software engineering environments (PSEE) in that we have made some effort to define some potential stakeholders who might have an interest in the type of environment that we are proposing. Based on our understanding of the tasks that they are required to undertake, and the difficulties associated with those tasks, we have postulated some priorities based on the stakeholder populations' needs. We suggest that the primary, long-term stakeholders who will benefit from the advantages that an SPTME can provide are software development teams and their managers. In the first part of the thesis we review the evolution of software engineering environments into the commercial environments that are in use today, and then compare and contrast a number of process­ centred software engineering environments. Drawing on open PSEE research issues that have already been identified by leaders in the field, we select a subset that are particularly relevant to our work to define the specific objectives for this work. To provide the foundation for our work with SPM, we propose the software engineering process standard (IS 15504, SPICE) as a source of abstract processes for deriving a concrete process model to use to introduce specified, repeatable, software development processes into a live, software development group. In the second part of the thesis we firstly develop the conceptual basis of the Dukas Method (OM) and its implementation as a hierarchical software process modelling method that captures attributes of an existing or proposed software process from behavioural, inormation and temporal viewpoints. There is a brief f discussion of how we have extended a commercial drawing package to capture and export information about OM models. We describe how we have used the Dukas Method to model a set of software processes, derived from the SPICE reference framework, for our development group. In later chapters of this thesis we describe some of the underlying architectural and design issues that were of concern during the development our proof-of-concept prototype Dukas Environment (DE) that implements and enacts descriptions of software process models. This environment has, as its core component, a process engine that enacts tasks according to either temporal or informational stienvironment or to a member of the development team. As a result of our efforts in integrating commercial tools into the environment we are able to articulate a set of attributes with which to assess future tools for their suitability for integration, and to integrate them using widely available behaviour-invocation technologies. We describe important aspects of the current prototype from the perspective of team members and project managers who are representative of our stakeholder populations. We conclude the thesis with a discussion of the results of our research in the context of its original gaols and our experiences using both the method and environment with our software development team. We look, briefly, at the relationship between our results and the work of other researchers in the field, and conclude with some proposals relating to future directions that this work might take