Which COBOL data type requires the most careful mapping when generating Java, and why?

Think of it like working hands-on with IBM mainframe systems — the answer here is 'PIC X alphanumeric fields, because Java has no fixed-length String type and padding must be handled explicitly'. PIC X fields in COBOL are fixed-length, space-padded character strings. Java Strings are variable-length, so the generated code must handle padding and trimming explicitly to preserve field-level comp. On real mainframe exams, this concept comes up repeatedly so make sure you have it locked in.