LIMPET(ID:4042/lim006)

Subset of LIMP 

Subset of LIMP

Hunt, William and Mary College 1979
Related languages
LIMP => LIMPET   Subset
References:
  • Hunt, J. et al, "Messages in Typed Languages" view details Abstract: Messages are increasingly being used for interprocess communication. The problem of introducing messages into typed languages is considered, and a solution in terms of typed message-channels is presented. Our particular treatment permits dynamic connexions, including secure linking of separately-compiled programmes, and also features nondeterminacy, thereby enabling automatic resource-scheduling without monitors. Implementation considerations are discussed, and a comparison with the work of other authors is given.
    Extract: Intro
    Since the construction of the first real-time systems, the notion of processes as a means of structuring programmes has become well-established. Early "real-time" languages (e.g. Coral 66 and RTL/2} concentrated primarily on generating "efficient" (i.e. fast) code; whilst later languages (e.g. Pearl, Procol, Progress) included constructs for process management, scheduling, communication and synchronisation. Traditionally, communication is by means of shared variables or data areas: and countless papers have been written on various methods of achieving synchronisation so as to reduce the frequency of errors. Thus more recently we have Concurrent Pascal and Modula in which monitors have been introduced to automate some of the synchronisation.

    With the advent of distributed systems, more attention has been devoted to killing the birds Communication and Synchronisation with the stone of message-passing. This concept, if not palaeolithic, is not new: but hitherto the method has often been disregarded, owing to the relative slowness of the resulting code. Jackson /18/, for example, advocates the technique for the design of systems, but not for the actual implementation. However, this situation is changing. Messages have been infiltrating operating systems for some years, e.g. [...] and a high-level language (EPL) has been produced for the implementation of systems of processes communicating by means of messages.

    EPL is based on BCPL, and is therefore typeless. Typeless languages are unable to provide the security of typed languages, and so will not be considered suitable for many applications. If messages are to be more-widely used, it therefore seems desirable to consider how they may be incorporated into typed languages. This is the problem we address in this paper.
          in SIGPLAN Notices 14(01) January 1979 view details