Raju Halder

This paper describes experiments with the development of a parallel computing platform on top of a compatible C# implementation: the Mono project. This implementation has the advantage of running on both Windows and UNIX platforms and has reached a stable state. This paper presents performance results obtained and compares these results with implementations in Java/RMI. The results show that the Mono network performance, critical for parallel applications, has greatly improved in recent releases, that it is superior to the Java RMI and is close to the performance of the new Java nio package. The Mono virtual machine is not yet so highly tuned as the Sun JVM and Thread scheduling needs to be improved. Overall, this platform is a new alternative to explore in the future for parallel computing.

Replicated services are often required to sustain high loads of multiple concurrent requests. This requirement is hard to balance with strong consistency. Typically, to ensure inter-replica consistency, all replicas should receive all updates. Unfortunately, in this case, a single slow replica may degrade the performance of the whole system. This paper proposes a novel reliable broadcast primitive that uses semantic knowledge to weaken reliable delivery guarantees while, at the same time, ensuring strong consistency at the semantic level. By allowing some obsolete messages to be dropped, the protocol that implements this primitive is able to sustain a higher throughput than a fully reliable broadcast protocol. The usefulness of the primitive and the performance of the protocol are illustrated through a concrete example.

Current data abstraction mechanisms are not adequate to control sharing of state in the general case involving objects in linked structures. The pervading possibility of sharing is a source of errors and an obstacle to language implementation techniques.
We present a general extension to programming languages which makes the ability to share state a first class property of a data type, resolving a long-standing flaw in existing data abstraction mechanisms.
Balloon types enforce a strong form of encapsulation: no state reachable (directly or transitively) by a balloon object is referenced by any external object. Syntactic simplicity is achieved by relying on a non-trivial static analysis as the checking mechanism.
Balloon types are applicable in a wide range of areas such as program transformation, memory management and distributed systems. They are the key to obtaining self-contained composite objects, truly opaque data abstractions and value types—important concepts for the development of large scale, provably correct programs.

n/a

The Robot Operating System (ROS) is one of the most popular open source robotic frameworks, and has contributed significantly to the fast development of robotics. Even though ROS provides many ready-made components, a robotic system is inherently complex, in particular regarding the architecture and orchestration of such components. Availability and analysis of a system’s architecture at compile time is fundamental to ease comprehension and development of higher-quality software. However, ROS developers have to overcome this complexity relying mostly on testing and runtime visualisers. This work aims to enhance static-time support by proposing, firstly, a metamodel to describe the software architecture of ROS systems (the ROS Computation
Graph) and, secondly, model extraction and visualisation tools for such architectural models. The provided tools allow users to specify custom-made queries over these models, enabling the static verification of relevant properties that had to be (manually) checked at runtime before.

VirtuosoNext{\$}{\$}^{\{}{\backslash}text {\{}TM{\}}{\}}{\$}{\$}is a distributed real-time operating system (RTOS) featuring a generic programming model dubbed Interacting Entities. This paper focuses on these interactions, implemented as so-called Hubs. Hubs act as synchronisation and communication mechanisms between the application tasks and implement the services provided by the kernel as a kind of Guarded Protected Action with a well defined semantics. While the kernel provides the most basic services, each carefully designed, tested and optimised, tasks are limited to this handful of basic hubs, leaving the development of more complex synchronization and communication mechanisms up to application specific implementations. In this work we investigate how to support a programming paradigm to compositionally build new services, using notions borrowed from the Reo coordination language, and relieving tasks from coordination aspects while delegating them to the hubs. We formalise the semantics of hubs using an automata model, identify the behaviour of existing hubs, and propose an approach to build new hubs by composing simpler ones. We also provide tools and methods to analyse and simplify hubs under our automata interpretation. In a first experiment several hub interactions are combined into a single more complex hub, which raises the level of abstraction and contributes to a higher productivity for the programmer. Finally, we investigate the impact on the performance by comparing different implementations on an embedded board.