Although the number of proposals discussing various atomic action schemes is increasing, these schemes are very rarely used in designing practical applications. To a large extent, this is accounted for by the gap existing between the languages used in research and the standard or widely spread languages (e.g. C, C++, Ada 83, Ada 95, Java) employed by practitioners. Moreover, very often researchers extend languages with new features or invent new languages to express their ideas better. Even though these approaches seem to be quite natural, they widen the gap between practice and research. To bridge this gap, we should consider fault tolerance schemes in terms of a standard language, taking the language itself for granted. The question which we believe should be addressed is how to use/implement a particular scheme in these languages rather than how to modify the language. Only in this way the schemes could be used directly and the application domains of atomic action schemes extended. The main intention of this paper is to summarise research that has been done in the last years in designing various atomic action and

this paper is to offer a general approach to designing distributed Coordinated Atomic action schemes. Problems of action components partitioning and distribution are discussed. We consider ways of dealing with external and local objects within distributed Coordinated Atomic action schemes; several proposals are discussed in detail. The approach proposed relies on using forward error recovery in the form of distributed and concurrent exception handling and resolution. After discussing the general approach, we demonstrate how it can be applied when the standard distributed model of Ada 95 is used. The presentation of the scheme is sufficiently detailed for it to be used in practice. In particular, a thorough description of the action support and all patterns (skeletons) required for designing application software are given. 1. Coordinated Atomic Actions 1.1. Atomic Transactions and Conversations

This paper proposes a new N-version programming (NVP) scheme which allows several caller tasks to jointly use components which are designed diversely. Diversity is applied here at the level of classes in such a way that several version classes (objects) are developed separately and independently, and are encapsulated into a diversely designed object. Such objects are to be implemented in a special stylised way to incorporate a controlling mechanism which would deal with task and version synchronisation, adjudication of version output parameters and states, faulty version recovery, etc. The general approach is demonstrated using Ada. We outline the characteristics of applications which benefit from using such NVP scheme, discuss the engineering of diversely designed objects and of the software which uses them and describe several possible extensions of the scheme. Keywords: software design diversity, N-version programming, object-orientation, re-use, Ada