SPA2005 session: The SmartReconciliation project at BNP Paribas

One-line description:Lessons learned while developing a high performance, flexible reconciliation engine using objects and arrays.
 
Session format: Case study [read about the different session types]
 
Abstract:This session describes the lessons learned when developing the SmartReconciliation engine at BNP Paribas.

The SmartReconciliation architecture combines Java components with a functionally-driven array engine. In the initial design, much of the array implementation leaked out into other layers. This led to several extensive refactorings.

Within the application reconciliations are specified using a domain-specific language, which is executed by an implementation of the GOF interpreter pattern. The parser for the language is generated by SableCC.

Because the requirements change frequently, we have had to develop techniques for generating the grammar and compiler automatically from the object model.

 
Audience background:To get the most from the session you will have had experience of OO development and design. It will help if you have have a basic understanding of domain-specific languages, BNF and code generation.

 
Benefits of participating:Find out about the advantages and pitfalls of an architecture that mixes OO and functional/array approaches.

Hear about the issues in applying a domain-specific language in an evolutionary environment.

See how model-driven language generation can reduce the impact of change.
 
Materials provided:Slides and pdf of case study presentation.
 
Process:The slide presentation will be punctuated by two group exercises. It will be followed by a Q&A session.

 
Outputs:Web page including Q&A results and presentation as pdf.
 
History:Some of the material on model-driven language generation will be presentated at the SPA MDA day in October 2004.

 
Presenters
1. Romilly Cocking
Cocking and Co. Ltd.
2. 3.