Re: Compiler directive

Steve,

either method looks confusing.  Do people use /if defined to bring in code

or prototypes?

Personally, the only in-line code I bring in using an /IF DEFINED is debug
code. But then, the only in-line code I bring in using a /COPY is 'standard'
*PSSR code - I don't like using compiler directives (any of 'em!) in
C-specs.

Instead of conditionally compiling code I prefer to have the code exist in

service program procedures.  And instead of conditionally including
prototypes it is easier for me to have one PR source member for each service
program. Then the module being compiled simply includes the PR source
members of the service programs it will be calling into.  I find that even
5000 line PR source members compile fairly quickly.

What do you do about duplication? For instance, what do you do if you have a
program which calls 2 IBM API's, both of which use QUSEC? Do you have a
copybook which defines the QUSEC structure and explicitly include it in your
program? Do you have any 'standard' variables that are used by different
procedures in different service programs?

I tend to use one copybook per module - each module tends to include related
procedures, so it makes sense to group them that way. In a multi-module
*SRVPGM, this might mean that an application program would
include several /COPY's for each module used instead of a single
*SRVPGM-level /COPY, but I think the reduction in needless procedure
definitions is worth it.

...bottom line is I recommend against conditional compilation.

Fair enough. I think we would probably agree that it makes sense to have a
single copybook per 'group' of related functions, whether it be at a module-
or *SRVPGM-level.

Rory