COBOL REPLACE Overlap Effects on Java Interoperability

This question addresses the runtime behavior of the COBOL REPLACE statement when its target field overlaps with another field in the same record, within a COBOL-to-Java interoperability context. The COBOL language standard does not define the outcome when REPLACE modifies overlapping fields, resulting in compiler-dependent results. IBM Enterprise COBOL and Micro Focus COBOL may apply different overwrite or merge semantics, which can lead to unexpected data values when COBOL records are exposed to Java consumers. This ambiguity impacts the reliability of data contracts that assume consistent field modification behavior across platforms.

Under what conditions, if any, do compilers define consistent overlapping-field behavior? Can a portable interface be designed without relying on REPLACE overlap semantics? What mitigation strategies exist for ensuring defined data exposure in mixed-COBOL-Java deployments?