Distributed PS-algol

Abstract

Fundamental to the concept of persistence is locality transparency whereby a programmer needs not be aware of the locality of data. It is argued that the locality transparency principle can be generalized in the context of a distributed environment. The implication is that syntax and semantics in the manipulation of remote data is the same as for local data. Distributed programming is then no more difficult than conventional persistent programming. But because of the locality transparency principle, many advantages offered in a distributed environment become inaccessible.

This paper presents a language design which on one hand supports persistent programming while on the other supports the kind of coding that requires the whereabout of data to be discovered; a language concept locality is introduced for such purposes. Other features of the language are presented: lightweight processes that may be made to run on a processor different from the one it is defined, a remote procedure call mechanism that supports call-by-value parameter passing semantics, dynamic typechecking with a hint of polymorphism of separately compiled communicating processes, and semaphores with syntactic support for common usages.