Michael M. Gorlick

Using weaves for software construction and analysis

The authors discuss the architectural features of weaves, their implementation, and their use in a variety of applications. Weaves are networks of concurrently executing tool fragments that communicate by passing objects. Weaves are distinguished from other dataflow styles by their emphasis on instrumentation, continuous observability, and dynamic rearrangement: basic low-overhead instrumentation is inserted automatically, executing weaves can be observed at any time by means of sophisticated analysis agents, without degrading the performance of the weave, and weaves can be dynamically snipped and spliced without interrupting the data flow.<<ETX>>

Real-time interval logic for reasoning about executions of real-time programs

Research on the testing and debugging of distributed real-time programs now focuses on more formal approaches to specification and testing. Temporal logic is a natural candidate for this since it can specify properties of event and state sequences. However, the absence of any concept of real-time limits the application of temporal logic to non real-time behavior. This paper presents an extension of the interval logic of Schwartz et al. [SMSVP83], by increasing the expressive power of the logic (with respect to real time) while retaining its intuitive appeal and understandability. The extensions are added in a “layer” that can be stripped away if formal verification is the goal, or retained if timing behavior must be tested. The extensions include: the ability to deal with real time (as in [JM86b, JM86a, OW87, NA88]); more powerful interval specification mechanisms; a limited form of quantification; and the direct expression of event predicates (as in [LeD86]). Since our work is intended to yield practical tools for software testers, we emphasize the ease of expressing the complex timing properties of real-time software (e.g. periodic behavior, performance constraints), and we demonstrate the use of the interval logic on some real-time examples that represent a test of the expressiveness and understandability of the notation.

Mockingbird: a logical methodology for testing

Abstract Mockingbird is a testing methodology founded on a formal specification of the test space. The specification is executable and bidirectional. When run in one direction it acts as a generator, producing tests whose properties conform to the specification. When run in the opposite direction it acts as an acceptor, validating tests against the specification. The specification language is a combination of context-free grammars and constraint systems. The semantics of the specification are based on constraint logic programming. This paper describes the philosophy, design, and implementation of Mockingbird and its use in testing a large, complex system.

Raging incrementalism: harnessing change with open-source software

Change is a bitter fact of life for system developers and, to a large extent, conventional practices are aimed at arresting change and minimizing its effects. We take the opposite view and are exploring system engineering practices that harness the forces of change for the ongoing, incremental improvement of systems---a view we name raging incrementalism. We harness three powerful forces to ride the waves of change: open-source software, commodity hardware, and web-like, representational state transfer architectures. This paper describes an early experiment in applying raging incrementalism to a complex system: large-scale digital video capture, distribution, and archival for launch range operations. We outline the methodology of raging incrementalism, describe the vital role open-source plays in system development and construction, and offer insights on the programmatic consequences of embracing open-source software.