RE: R: Library and Source File Question

I'll weigh in.

For smaller, self-contained applications, I keep data and programs together,
but for larger environments, I separate source, data and objects.  The
latter harkens back to my SSA days.  It allows multiple data sets with the
same programs, which can come in handy.  Of course, your application has to
be written that way (basically, you need a way to determine which data
library you're in), but in general its a good approach.

As to source naming, I've migrated to a two-file system, QSOURCE and
QMODSRC.  QMODSRC contains modules, while QSOURCE contains eevrything else.
I've created my own compile command that uses the source type to do the
appropriate compile command, but it also checks the source file name to
determine whether it should be creating a module or a program.

The toughest issue with my source file approach is dealing with commands and
programs; since they have the same source file, they need different names.
And since the default for CRTCMD is to use the name of the command as the
name of the CPP, you can't use the defaults.  Unfortunately, there's no way
to override this in the source (like you can override certain compile
characteristics with an H spec in RPG).  Therefore, my compile command
defaults the CPP name to the command name plus a "C".  This limits my
command names to nine characters, but seems to be a reasonable compromise.

Joe