Abstract

Persistent programming systems are generally recognised as the appropriate technology for the construction and maintenance of large, long-lived application systems. Many successful prototypes have been constructed, and a large body of application building experience is emerging. However, all persistent systems to date have been provided within an operating system environment, and the services provided by the operating system have been relied upon to give the necessary support for program tasks such as editing and linking. Only the execution of persistent programs has occurred within the persistent environment. Here we examine some of the consequences of providing all the support required for program construction within the same environment as the persistent data. We show how having the construction, compilation, and execution processes all operating within a single environment leads to some powerful new techniques. This new power is achieved by the sharing of persistent data across the hitherto enclosed boundaries of these processes.

Citations