DMS. Deterministic code transformation at an industrial scale.

Collect and Parse Source Code

Legacy repositories are often fragmented, inconsistent, and partially broken.

DMS:

• Collects and parses all source components targeted for transformation
• Identifies and reports anomalies such as:
  • Code that no longer compiles
  • Obsolete or dead code
  • Missing copybooks
  • Structural inconsistencies
• Parses not only COBOL, but also related technologies that must be transformed — including embedded SQL and database access layers

This creates a complete and analyzable model of the system before transformation begins.

System Analysis

DMS analyzes the system at multiple levels

Compilation Unit Level

• Control-flow analysis (e.g., GOTO elimination)
• Data-flow analysis to preserve the precise data memory model
• Accurate handling of COBOL storage semantics and backing bytes

Global System Level

• System-wide refactorings
• Copybook consolidation into modern class structures
• Global identifier renaming
• Cross-program dependency analysis

This multi-level analysis enables safe structural transformation at scale.

Apply Rewrite Rules

DMS rewrite rules are written using the syntax of both the legacy and target languages, making them readable and debuggable by engineers.

These rules are:

• Conditional
• Composable
• Context-aware
• Driven by analysis results

This allows DMS to perform complex, semantically accurate transformations while improving maintainability of the resulting code.

If DMS encounters a construct it cannot confidently translate, it flags it for engineering review. SD engineers can then extend the rewrite rule set to handle the special case. This iterative process continues until:

• 100% of the code is translated
• The entire system compiles successfully

Apply Global Optimization Rules

After 100% translation is achieved, additional rewrite rules can be applied to optimize the architecture according to customer objectives.

Examples include:

• Separation of data access into a dedicated data layer
• Dead code removal
• Consolidation of redundant structures
• Migration toward modern frameworks (e.g., Spring Boot architecture)

These transformations move beyond translation toward modernization.

Pretty Printing

DMS includes configurable pretty-printing capabilities.

Output formatting can be aligned with the customer’s coding standards, ensuring:

• Readable, maintainable code
• Consistent style guidelines
• Clean, production-ready output

Iterate with Customer Feedback

Customer engineers review the translated code and often propose further refinements.

Because the process is fully automated and rule-driven:

• SD can adjust rewrite rules
• Regenerate the system
• Deliver consistent improvements across the entire codebase

Changes are not manual edits — they are systematic enhancements applied uniformly.

Regression Testing and Validation

SD maintains and executes thousands of DMS translation regression tests to validate transformation accuracy.

This provides:

• High confidence in semantic correctness
• Early detection of potential issues
• Reduced downstream functional testing effort
• Lower overall modernization risk

By identifying transformation issues before system-level functional and performance testing, DMS significantly shortens the path to deployment.