10th International Software
Product Line Conference
(SPLC 2006)
21-24 August 2006
Baltimore,
Maryland, USA
Keynote Speakers
Carliss Baldwin, Harvard Business School - Unmanageable Design Architectures: What They Are and Their Financial Consequences
Gregor Kiczales, University of British Columbia - Radical Research In Modularity: Aspect-Oriented Programming and Other Ideas
Unmanageable
Design Architectures: What They Are and Their Financial Consequences
by
Carliss Y. Baldwin
Behind every innovation lies a new design. Large or complex designs, involving many people, require architectures that create a sensible subdivision of the design tasks.
Design architectures (and the systems built from them) may be "manageable" or "unmanageable." By manageable, I mean that the artifacts created within the architecture will stay within the boundaries of a single enterprise (or a supply chain controlled by a dominant firm). Windows and Office are manageable architectures by this definition, whereas Apache and Linux are coordinated but not manageable. "Manageable" architectures give rise to product lines and product families, while "unmanageable" architectures give rise to modular clusters and open source communities.
There are important technical properties of a design architecture that affect its manageability. In this speech, I will talk about how designs draw resources from the economy, and what technical properties make an architecture "manageable" or "unmanageable." These properties, I will argue, are not good or bad in themselves, but they affect economic incentives and patterns of competition over new products and designs. Thus design architecture is an important consideration in formulating a sound product line strategy.
Radical
Research In Modularity: Aspect-Oriented Programming and Other Ideas
by
Gregor Kiczales
Modularity is a motherhood principle in our field. Just as politicians love to kiss babies for the camera, computer scientists love to preach the virtues of good modularity.
But what does modularity mean? In our field the idea has typically been equated with a notion of cellular or even block structure, where each block or module defines its interface with the surrounding modules. A close examination suggests this notion is too restrictive: it fails to support construction of complex systems, it fails to account for practice, and it fails even to be intuitively satisfying.
Work in biology and other fields suggests that there are many other possible kinds of modularity, and recent research in aspect-oriented programming shows what some new forms of modularity for software might be. Building on this we outline a line of attack for discovering other kinds of modularity and making productive use of new kinds of modularity in building real systems.
