Persistent object systems

Persistent object systems address the challenging requirements of long-lived, data-intensive applications. At their most ambitious, they aim to support simple, reliable and efficient access to large, shared bodies of structured data and programs, over extended periods of time. Such aims lie at the nexus of an ongoing convergence between historically distinct communities. On the one hand the research community has concentrated its focus on orthogonal persistence, a programming language abstraction in which the longevity of data, including programs, is independent of the way in which it is created and manipulated. As such, orthogonal persistence integrates the database view of longevity of information with the programming language view of data abstraction and manipulation, all within a coherent system. Meanwhile, fuelled by commercial interest in object-based systems, industry has worked to bring objects into mainstream products. Whether the context is storage and manipulation of Java or C++ objects, or support by vendors for complex data models in pure object or hybrid objectrelational databases, the management of long-lived structured objects is now an important commercial opportunity. The fusion of the persistence abstraction with database functionality such as transactions and crash resilience promises increased programmer productivity, application safety, and data integrity. The recent popularity of application programming languages such as Java that emphasise safety over raw efficiency will ensure continued interest in persistent object systems, as applications demand clean and efficient persistence solutions for the objects they create. In this exciting context, we are privileged to bring this special issue on Persistent object systems to SP&E, comprising papers that run the gamut from low-level operating system support through language design and implementation, experimental evaluation of prototypes and evaluation methodologies, on up to high-level software architecture.We received 25 submissions, accepting six for this special issue, having observed the standard SP&E process of peer review. Two other submissions were recommended for separate publication in regular issues of SP&E. Authors of several of the remaining submissions were encouraged to revise and resubmit their papers for similar consideration. The first paper, Operating system support for persistent systems: past, present and future, by Dearle and Hulse, surveys the history of explicit operating system support for the persistence abstraction, gives consideration to modern trends in operating systems, and expounds the philosophy and design of a new ‘nano-kernel’ operating system called Charm, whose explicit goals are flexible support of persistent applications via a minimal set of system primitives.