Re: QCLSRC, QRPGSRC, etc... best practice?

On 10/29/2018 1:15 PM, Booth Martin wrote:

This leads to the issue I am having.  I was taught early on that
hard-wiring names into programs was a Red Flag and a promise for future
nightmares of odd errors.

Hard coding is indeed a red flag, but some are unavoidable with the
current implementation.

My own work practice is that I have a source library of QDEV which has
my source member models for everything from subfiles to message files to
sql to filters to  on&on.  Current projects are also in there.  One
advantage is that this guarantees all objects will have unique names
within the project.  Also, when a project is ready to be handed off
everything is all there in 1 place.  Forgetting pieces disappears as a
problem.   /COPY in that scenario is something I have never figured out,
but now it appears that prejudice must be buried with MOVE, GOTO, and
VARPG.

"... when specifying a physical file, the library, file, and member
name, can be in one of these formats:

|libraryname/filename,membername filename,membername membername |

 * A member name must be specified.
 * If a file name is not specified, QRPGLESRC is assumed. ..."

Well, you can skip the library; the compiler will find the source file
in the library list. You can specify the RPG source file as whatever you
like to use, for example

/copy myrpgsrc,mycpymbr
/copy qprotosrc,mycpymbr
/copy qcpysrc,mycpymbr

If you really want all the pieces/parts in a single source file named
mysrc, then you might have something like this

mysrc custsrv - Customer service program
mysrc custbnd - Customer srvpgm binder
mysrc custcpy - Customer srvpgm prototypes
mysrc custinqr - Customer inquiry main
mysrc custdsp - Customer inquiry dspf
mysrc custpf - Customer pf
mysrc custl1 - Customer lf by name
mysrc custmake - Build the Customer application from source

Inside mysrc custinqr, you'd have
/copy mysrc,custcpy
...
largeCustInd = isLargeCust(customerID);
...

So yes, there is some hard coding, but it's a minimum amount of yuck :-)

Scott, it appears that IBM's preferred choice is not QCPYSRC, either. 
My own first-impression preference would be a default of *FILE.

Pop in an RFE; that's an eminently useful suggestion. If you're in a
hurry, consider modifying one of the available preprocessors to do this
for you.

Anyway, that pretty well sums up my quandary.  It looks to me like using
the default probably works best, but I better listen to the people who
have already been there.

Not sure why the default has any particular value other than lots and
lots of people... er, default to it. :-/