SoftwareMining •
COBOL to Java & C#
• Predictable & Repeatable
COBOL Business Rule Extraction (BRE) attempts to identify and isolate the code responsible for business decisions, such as eligibility, pricing, validation and calculation rules. This can be challenging because business logic is frequently interwoven with transaction processing, database access, file handling and screen management.
This guide examines the benefits and limitations of extracting business rules directly from COBOL and explains how complete application translation can provide a safer basis for understanding, preserving and selectively refactoring those rules.
Consider a hypothetical example in which an organization needs to identify the logic used to calculate customer discounts. A transaction program may contain thousands of lines covering screen handling, database access, validation, error processing and calculations. Only part of that code may represent the discount rules, but those rules can depend on data and processing elsewhere in the application.
BRE tools may use static analysis, dependency tracing and heuristics to identify potentially relevant code. However, extracted rules normally require review because important dependencies can extend across programs, copybooks, files, database tables and external services. Removing logic from its surrounding processing context can also make its original behaviour more difficult to interpret.
SoftwareMining takes a different approach by translating the complete application into structured Java or C# while preserving the relationships between its business logic, data structures and supporting operations. The generated source can then be examined using familiar development tools without first attempting to separate business rules from the rest of the application.
Large COBOL applications commonly combine reporting, transaction processing, database and file access, screen management, validation and business decisions. Only some of this functionality is suitable for implementation in a business-rules engine or separate rules service. SoftwareMining supplements the generated Java or C# with application analysis, dependency information, call-flow diagrams and documentation .
One modernization strategy is to translate the complete application into maintainable Java or C# before deciding which business rules should be separated. This preserves the working application context while allowing suitable modules, such as eligibility, pricing or calculation logic, to be identified and selectively refactored into a rule-based architecture.
This staged approach reduces the need to identify every business rule correctly before modernization can begin. It also allows extracted rules to be compared with the behaviour of the translated application during testing.
After translation, development environments such as Eclipse IDE and Microsoft Visual Studio provide facilities including reference searches, call hierarchies, navigation between classes and debugging. These facilities help developers trace data usage, follow program flow and understand how translated business logic interacts with the rest of the application.
SoftwareMining supplements the generated Java or C# with application analysis, dependency information, call-flow diagrams and documentation. Together, these capabilities provide a navigable representation of the application that can be used to understand business logic and plan subsequent refactoring.
Organizations can therefore use translation to preserve the complete application first, and then isolate selected business rules where doing so provides a clear architectural or operational benefit.