The Problem. Computing technology now affects nearly every dimension of modern society: managing critical infrastructure such as power grids and telecommunication networks; supporting electronic commerce and medical information systems; and controlling the operation of aircraft and automobiles. This pervasiveness of computing technology, coupled with its rapidly increasing complexity, gives rise to the need for computer systems that are able to adapt to changing conditions. In the last decade, extensive research has been conducted on many aspects of self-adaptive software systems. Examples include adaptive software mechanisms [1 20]; software-architecture-based techniques for supporting dynamic adaptation [21 38]; adaptable and extensible operating systems [39 42]; and requirements-level and formal methods-based techniques [43 52]. This research has greatly improved our understanding of adaptive software and several key supporting concepts, including computational reflection [53 55], separation of concerns [12, 56], component-based design [57, 58], and transparent interception of program flow [59 61]. Despite these advances, designing an adaptive software system remains a very challenging task. We speculate that much of the difficulty is due to the fact that adaptive software is being designed and implemented using traditional tools and environments intended for the development of non-adaptive software. We contend that the full potential of dynamically adaptive software systems cannot be realized without fundamental advances in the corresponding development environments. Such environments must enable developers to explicitly address those aspects of the design problem that distinguish adaptive systems from non-adaptive systems. These issues include anticipating how the software may need to adapt in the future, constructing decision-making software to govern the adaptation, and ensuring that system integrity is not compromised by adaptation.
