The project aims to promote the Behaviour Description Language (BDL) as a back-bone formalism, linking together the dynamic views of the Unified Modeling Language (UML). The BDL project is part of the Reutel project, a joint project with Alcatel.
Researchers and postgraduate students contributing to BDL are members of the PAMPA and EP-ATR projects of IRISA.
We introduce the definition of a language of distributed reactive objects, a Behaviour Description Language (BDL), as a unified medium for specifying, verifying, compiling and validating object-oriented, distributed reactive systems. One of the novelties in BDL is its seamless integration into the Unified Modeling Language approach (UML). BDL supports a description of objects interaction which respects both the functional architecture of system designs and the declarative style of diagram descriptions. This support is implemented by means of a partial-order theoretical framework. This framework allows to specify both the causality and the control models of object interactions independently of any hypothesis on the actual configuration of the system. Given the description of such a configuration, the use of BDL offers new perspectives for a flexible verification of systems by modeling them as an asynchronous network of synchronous components. It allows an optimized code generation by using compilation techniques developed for synchronous languages. It permits an accurate validation and test of applications by supporting the manipulation of both causal and control dependencies. BDL aims at maximizing the re-usability of high-level specifications while minimizing programming effort and test-case based validation of distributed systems.
J.-P. Talpin, A. Benveniste, B. Caillaud, C. Jard, Z. Bouziane, H. Canon, in proceedings of ISORC'98, The 1st IEEE International Symposium on Object-oriented Real-time Distributed Computing, Kyoto, Japan, 1998.
We introduce the definition of a language of distributed reactive objects, a Behaviour Description Language (BDL), as a unified medium for specifying, verifying, compiling and validating object-oriented, distributed reactive systems. One of the novelties in BDL is its seamless integration into the Unified Modeling Language approach (UML). BDL supports a description of objects interaction which respects both the functional architecture of system designs and the declarative style of diagram descriptions. This support is implemented by means of a partial-order theoretical framework. This framework allows to specify both the causality and the control models of object interactions independently of any hypothesis on the actual configuration of the system. Given the description of such a configuration, the use of BDL offers new perspectives for a flexible verification of systems by modeling them as an asynchronous network of synchronous components. It allows an optimized code generation by using compilation techniques developed for synchronous languages. It permits an accurate validation and test of applications by supporting the manipulation of both causal and control dependencies. BDL aims at maximizing the re-usability of high-level specifications while minimizing programming effort and test-case based validation of distributed systems.
A. Benveniste, B. Caillaud, P. Le Guernic, in proceedings of CONCUR'99, Concurrency Theory, 10th International Conference, Eindhoven, The Netherlands, August 1999, LNCS 1664, Springer.
We present an in-depth discussion of the relationships between synchrony and asynchrony. Simple models of both paradigms are presented, and we state theorems which guarantee correct desynchronization, meaning that the original synchronous semantics can be reconstructed from the result of this desynchronization. Theorems are given for both the desynchronization of single synchronous programs, and for networks of synchronous programs to be implemented using asynchronous communication. Assumptions for these theorems correspond to proof obligations that can be checked on the original synchronous designs. If the corresponding conditions are not satisfied, suitable synchronous mini-programs which will ensure correct desynchronization can be composed with the original ones. This can be seen as a systematic way to generate ``correct protocols'' for the asynchronous distribution of synchronous designs. The whole approach has been implemented, in the framework of the SACRES project, within the Sildex tool marketed by TNI, as well as in the Signal compiler.
presented by Benoît Caillaud
presented by Jean-Pierre Talpin
Maintained by Benoît Caillaud <Benoit.Caillaud@irisa.fr>, updated $Date$, copyright IRISA