MFED vs native z/OS compile differences

[Abhishek Gupta]

Has anyone here faced something similar or have a suggestion? Compiles on MFED but fails on z/OS. What differences should we check first? Thanks in advance for any pointers.

[Anuj Dhawan]

MFED uses the Micro Focus COBOL compiler, while z/OS normally uses IBM Enterprise COBOL. MFED provides strong mainframe compatibility, but the two compilers are not identical. The first things to compare are:

• COBOL dialect: MFED may accept Micro Focus extensions or tolerate syntax that IBM Enterprise COBOL rejects. Ensure MFED is compiling with the appropriate mainframe/IBM dialect and without unintended compatibility extensions.
• Compiler versions: Check the MFED version and the IBM Enterprise COBOL version. A language feature supported by MFED may not exist in the installed IBM compiler release.
• Compiler directives/options: Compare the MFED directives with the IBM compiler options, especially source format, reserved-word handling, ARITH[, TRUNC, NUMPROC, NSYMBOL, DYNAM, SQL and CICS settings.
• Copybooks: Confirm that both environments use exactly the same copybook versions. On z/OS, also verify the SYSLIB concatenation order.
• Source format: MFED can sometimes be more forgiving about tabs, columns, continuation lines and text beyond column 72. IBM fixed-format compilation may reject or truncate such lines.
• SQL/CICS processing: If the program contains EXEC SQL or EXEC CICS, confirm that the required precompiler, translator or integrated compiler option is active on z/OS.
• Character conversion: Check whether the source was transferred correctly from ASCII to EBCDIC and whether it contains hidden or unsupported characters.

The most important evidence will be the first IGY... error in the IBM compiler listing. Subsequent errors may only be cascading messages. If the first few IBM errors, compiler version and MFED directives are shared, the exact difference should be much easier to identify.

[hollow_rowan69]

When a program compiles under MFED but fails with IBM Enterprise COBOL, Anuj's list is the efficient order: dialect and compiler options first, then identical copybook versions and SYSLIB order, then source format past column 72. ARITH, TRUNC, NUMPROC, NSYMBOL and SQL or CICS precompile settings are where silent behavioral gaps hide even after the compile succeeds. The first IGY error in the IBM listing is the one to chase; later messages are often cascade noise. Confirming ASCII to EBCDIC transfer did not introduce hidden characters saves a long hunt through otherwise valid source.