A UTP semantic model for Orc language with execution status and fault handling

Abstract

The Orc language is a concurrency calculus proposed to study the orchestration patterns in service oriented computing. Its special features, such as high concurrency and asynchronism make it a brilliant subject for studying web applications that rely on web services. The conventional semantics for Orc does not contain the execution status of services so that a program cannot determine whether a service has terminated normally or halted with a failure after it published some results. It means that this kind of failure cannot be captured by the fault handler. Furthermore, such a semantic model cannot establish an order saying that a program is better if it fails less often. This paper employs UTP methods to propose a denotational semantic model for Orc that contains execution status information. A failure handling semantics is defined to recover a failure execution back to normal. A refinement order is defined to compare two systems based on their execution failures. Based on this order, a system that introduces a failure recovery mechanism is considered better than one without. An extended operational semantics is also proposed and proven to be equivalent to the denotational semantics.