We present a new approach to the control of versions of software and other hierarchically structured entities. Any part of a system, from the smallest component to a complete system, may exist in different versions. The set of all possible versions under the refinement relation forms a partial order (in fact, a lattice). The fact that version V approximates version V in this order means that V is relevant to V in this sense: when constructing version V of a system, we can sometimes use version V of a component if nothing more appropriate is available. More precisely, a particular version of an entire system is formed by combining the most relevant existing versions of the various components of the system. We call this the variant structure principle; it makes precise the idea that components of a given version of the system can be inherited by more refined versions of the system.

This paper describes the design of a configuration management toolkit, which integrates a dedicated version control system and shape, a significantly enhanced Make [7] program. Enhancements include full access to the version control system and support of configuration rules which control the selection process for component versions during identification, build or rebuild of system configurations. Since integration of version control was a major objective for our system, we had to face the need for a document identification scheme that goes beyond the usual way of specifying name and type of a document. As a consequence, we began to design an attributed filesystem (AFS) that introduces a much more generalized scheme for document identification. The AFS comprises concepts for version control, support for variants or status models for example. Furthermore, it helps to abstract from the particular underlying data storage system in such a way that it makes no difference whether it is an or...

The architecture is the fundamental framework for designing and implementing large scale software, and the ability to trace and control its evolution is essential. However, many traditional software configuration management tools view `software' merely as a set of files, not as an architecture. This introduces an unfortunate impedance mismatch between the design domain (architecture level) and configuration management domain (file level.) This paper presents a software configuration management model that allows tight version control and configuration management of the architecture of a software system. Essential features of the model have been implemented in a research prototype, Ragnarok. Two years of experience using Ragnarok in three, real, small- to medium sized, projects is reported. The conclusion is that the presented model is viable, feels `natural' for developers, and provides good support for handling an evolving architecture. 1. Introduction The software architecture is t...

After years using of Adele [3], a con guration management system, it became apparent that it lacks activity-related concepts and mechanisms like work environment control, user coordination and synchronization, method and tool control, etc. It was also clear that considerable work is required toadpat a Software Engineering Environment (SEE) to user requirements. Using this experience, Adele 2 has been implemented toprovide a general support for de ning and managing dynamic aspects of a SEE and facilitate the building of new SEE's. This paper describes, using an example (work space control), the concepts and mechanisms involved. We show how close integration of an activity manager with a software engineering database ful lls the basic requirements and how a high level task manager coupled toacon guration manager can be developed. Key words CASE; software development process; team coordination; programmingin-the-large; software engineering environment; process model; process control; groupware mechanisms.

For almost ten years, Make has been a most important tool for development and maintenance of software systems. Its general usefulness and the simple formalism of the Makefile made Make one of the most popular UNIX+tools. However,with the increased upcoming of software production environments, there is a growing awareness for the matter of softwareconfiguration management which unveiled a number of shortcomings of Make. Particularly the lack of support for version control and project organization imposed a hard limit on the suitability of Make for more complex development and maintenance applications. Recently,several programs have been developed to tackle some of the problems not sufficiently solved by Make. shape, the system described in this paper,integrates a sophisticated version control system with a significantly improved Make functionality, while retaining full upward compatibility with Makefiles. shape's procedure of identifying appropriate component versions that together fo...

Process-Centered Software Engineering Environments (PSEE) have recently attracted a large number of researchers. In such environments the software processes are explicitly described and interpreted by the PSEE, allowing software activities to be automated, assisted and enforced. [Lehman and Belady, 1985; Osterweil, 1987] claim that this capability is a central element in a software development environment for the improvement of software product quality and software developers productivity. We have addressed these problems in the framework of the Adele project. The Adele kernel, initially a configuration management system [Estublier et al., 1984], has been extended with respect to 1) modeling and support of complex product models: the Object Manager (OM) 2) modeling and support of software processes: the Activity Manager (AM), and 3) modeling and support of software product evolution: the Configuration Manager (CM). For data and product modelling, an ER/OO model has been implemented inc...