Traditional software lifecycles put significant effort in analysis, design, and implementation before their results can be tested. The start of testing late in the lifecycle often uncovers flaws from these early phases that require significant rework of analysis and design, causing cost overruns and schedule slippages. We describe a software lifecycle where each phase of development is followed by an execution test of the results of that phase. This lifecycle is progressive in the sense that it does not require significant "backing up" to earlier phases of development. The use of this lifecycle requires tool support that is just now becoming available. We examine this lifecycle and its implications for embedded software development. Section I examines traditional lifecycles. Section 2 describes the progressive lifecycle, explains its benefits and requirements, and briefly relates it to other software development paradigms. In conclusion, we relate development lifecycles to process maturity.

The Galois project at the University of Texas is building a programming environment that supports the formal development and verification of data structure programs. This programming environment supports features such as pointer manipulation and destructive update that often make formal treatment difficult.