IBM Enterprise COBOL V6R4+ - Add support for the equivalent of the -M option from the USS version of the compiler to the MVS version of the compiler

Hi,

The USS version of the IBM Enterprise COBOL V6R4 compiler, cob2, has a -M option that allows to produce a makefile of dependencies between the program and the copybooks it uses:

mkdepend.o: mkdepend.cbl
mkdepend.o: /home/usera/test/f171592/WSCOPY1.cpy
mkdepend.o: /home/usera/test/f171592/PDCOPY1.cpy
mkdepend.o: /home/usera/test/f171592/PDCOPY2.cpy

We would like to have for the MVS version of IBM Enterprise COBOL, a file output that lists the program dependencies, (SYSDEPS ?), containing some of the informations of type x'0060' record of the SYSADATA file, either formatted or parsable in a key/value format.

For example:

PROGRAM: MKDEPEND COPY: WSCOPY1 LIBRARY: USERA.TEST.F171592
PROGRAM: MKDEPEND COPY: PDCOPY1 LIBRARY: USERA.TEST.F171592
PROGRAM: MKDEPEND COPY: PDCOPY2 LIBRARY: USERA.TEST.F171592

However the purpose of this file is not to be used by the make utility, but only to know the dependencies of the compiled program.

This file could also contain other dependency information:

• Nested Programs, equivalent to record x'0046' in the SYSADATA file

• User Defined Functions, not tracked in the SYSADATA file?

This enhancement would allow us to exploit the list of copybooks actually used by the program, with the inclusion of nested copybooks, and the exclusion of copybooks not selected by conditional compilation, to build an automatic recompilation solution based on dependencies, or an impact study solution in case of copybook modifications.

For information, we use IBM Dependency Based Build in our compilation process, but IBM DBB does an analysis before compilation, based on scanners that do not process conditional compilation, (which incidentally produces an unreliable result), while the IBM Enterprise COBOL compiler is able to provide the list after compilation of the copybooks actually used.

Thanks.