Integrating Naming and Addressing of Persistent data in Programming Language and Operating System Contexts

There exist a number of desirable transparencies in distributed computing, viz., name transparency: having a uniform way of naming entities in the system, regardless of their type or physical make up; location transparency: having a uniform way of addressing entities, regardless of their physical location; representation transparency: having a uniform way of representing data, which simplifies sharing data between applications written in different highlevel languages and running on different hardware architectures (interoperability) and finally invocation transparency: having a uniform way of invoking operations on entities. The advent of persistency in programming language contexts has created a need for the integration of these four important concepts, viz., naming, addressing, representation and manipulation of data in programming language and operating system contexts. This paper attempts to address the first three transparencies, postponing the fourth to a later paper. First, we make up a list of things that are needed to construct a persistent programming environment and relate this list to existing persistent object models, revealing their inadequacies. We then describe a new model which merges programming language and operating system naming contexts into a global name space which, while enforcing uniformity through the use of globally unique names, still allows the application of personal nicknames. Furthermore, we explain how persistent data is stored and retrieved using a client/server model of interaction, and how it could be acted upon correctly, through the concept of typed data. We conclude by checking how well our model scores on the wish list, listing the current status and future directions for research.