One of the biggest challenges in breaking up your applications into multiple source members--and therefore into multiple modules--is setting up how to compile the source so that you end up with a program. This one activity, which can lead to enormous benefits in application maintenance, seems to keep many of us in the tired, old one-source-member-one-program model.
There are two very different ways you can create *PGM objects from multiple *MODULE objects. The first is to compile each source member independently of the other, using the CRTRPGMOD command, and then bind everything together with the CRTPGM command. The second is a less-complex method that can be accomplished by doing a little work up front.
Let's look at both methods.
Method 1
A typical program is made up of four modules. Each module is, of course, created from a corresponding source member. Therefore, there are four source members that need to be compiled into *MODULE objects so that the resulting program can be created.
To compile these source members, you have to first determine which one of the four is the so-called entry module. With RPG IV, the entry module is the source member that contains the mainline calcs and, consequently, the code to execute the RPG cycle at startup. The secondary source members, when compiled, will typically contain only subprocedures and perhaps some global variables and maybe a file or two, but not the RPG cycle startup code.
Once you've identified the entry module (it should be the only one with mainline calcs), you can compile all four of the source members to produce *MODULE objects. You use the CRTRPGMOD command to do that. As an example, assume that you're doing an Order Entry program that is made up of the following modules:
- ORDENTRY--The entry module, which contains mainline calcs and primary program logic
- FINDCUST-- Contains subprocedures that search the customer file(s)
- DATERTN--Contains subprocedures that perform date calculations
- ORDPRICE--Contains subprocedures that calculate pricing
If you look inside the source members that will be compiled to create these modules, you should see the Header specifications to support the compilation process, as follows:
ORDENTRY | .....H OPTION(*NODEBUGIO : *SRCSTMT) .....H/IF DEFINED(*CRTBNDRPG) .....H DFTACTGRP(*NO) ACTGRP(ORDENTRY) .....H/ENDIF |
FINDCUST DATERTN ORDPRICE | .....H OPTION(*NODEBUGIO : *SRCSTMT) |
The entry module has some specific keywords included in its Header specification. The OPTION keyword and its parameters aren't specific to compiling, but they do allow you to debug your programs more easily and should be included in most RPG IV source members. The DFTACTGRP(*NO) and ACTGRP(ORDENTRY) keywords are included only when the CRTBNDRPG command (PDM option 14) is used to compile the source member.
The secondary modules--FINDCUST, DATERTN, and ORDPRICE--have a simple Header specification as illustrated in the table above.
To compile these four source members, you can use PDM option 15 or a CRTRPGMOD command. The command to compile each of them would be as follows:
CRTRPGMOD MODULE(ORDENTRY) SRCMBR(ORDENTRY) +
SRCFILE(mylib/QRPGLESRC)
CRTRPGMOD MODULE(FINDCUST) SRCMBR(FINDCUST) +
SRCFILE(mylib/QRPGLESRC)
CRTRPGMOD MODULE(DATERTN) SRCMBR(DATERTN) +
SRCFILE(mylib/QRPGLESRC)
CRTRPGMOD MODULE(ORDPRICE) SRCMBR(ORDPRICE) +
SRCFILE(mylib/QRPGLESRC)
These four CRTRPGMOD commands will produce corresponding *MODULE objects for each of the source members (assuming there are no coding errors).
To create a *PGM object from these modules, you need to issue a CRTPGM command. On this command, you need to include the names of the modules that will be included in the *PGM program, as follows:
CRTPGM PGM(ORDENTRY) ACTGRP(ORDENTRY) +
MODULE(ORDENTRY FINDCUST DATERTN ORDPRICE)
On the MODULE parameter of the CRTPGM command, the CRTPGM command identifies the modules used to create the program. Note the order of the module names as listed on that parameter. The first module name is the name of the entry module and is the only one that impacts the CRTPGM command. If you want to use a different module as the entry module, you have to either specify that module name first or add the ENTMOD parameter to the CRTPGM command, as follows:
CRTPGM PGM(ORDENTRY) ACTGRP(ORDENTRY) ENTMOD(ORDENTRY) +
MODULE(DATERTN FINDCUST ORDENTRY ORDPRICE)
The ENTMOD keyword identifies the entry module that the CRTPGM uses when binding the modules together to create the *PGM object. When ENTMOD is used, the order of the module names on the MODULE parameter is irrelevant. If ENTMOD is not specified, the CRTPGM command defaults to ENTMOD(*FIRST).
Once the CRTPGM completes successfully, you have an executable program. In this case, you have the ORDENTRY program that is made up of the four modules.
The problem I have with this method of compiling is that you have to remember which modules are required when you're developing the application. This isn't too difficult if you are doing the development all in one day or you have written it down somewhere, like on a yellow notepad.
But suppose you have a program that is made up of not three or four modules, but 11 or 15. Try using the CRTPGM command to create that program, sign off, come back the next day, and continue the development. You have to enter all 11 or 15 module names on the MODULE parameter again. Needless to say, this isn't fun.
Method 2
The second method is a little easier and does not require that you even remember the names of the secondary modules. This technique allows you to "write" the names of the module in an OS/400 "notepad" and then refer to that notepad when compiling.
Of course, the notepad I'm talking about is a binding directory. Binding directories are simple, little objects on OS/400 that store the names of *MODULE and *SRVPGM (service program) objects. They store only the names of these objects, not the objects themselves.
Binding directories can be created anytime, and you can add to them the names of the modules and/or service programs anytime you want.
If you create a binding directory for each application you are creating and then add the names of the *MODULE objects to that binding directory, you can get away with a much easier compile process. For example, if you create a binding directory named ORDERBD and add to that binding directory the names of the secondary modules (FINDCUST, DATERTN, and ORDPRICE), you can avoid specifying those module names on the CRTPGM command.
To create a binding directory, use the CRTBNDDIR command as follows:
Then, add the names of the *MODULE objects to the binding directory. In this example, those names would be FINDCUST, DATERTN, and ORDPRICE. The ADDBNDDIRE command is used to add binding directory entries to a binding directory, as follows:
ADDBNDDIRE BNDDIR(ORDERBD) +
OBJ((FINDCUST *MODULE) (DATERTN *MODULE) +
(ORDPRICE *MODULE))
At this point, the binding directory is created, and it contains the names of the secondary modules. On the CRTPGM command, you can now use the binding directory instead of the long list of module names.
To create the ORDENTRY program, use the binding directory parameter in the CRTPGM command, as follows:
CRTPGM PGM(ORDENTRY) ACTGRP(ORDENTRY) +
MODULE(ORDENTRY) BNDDIR(ORDERBD)
As you can see, this technique makes the process a little easier. And when you sign off and come back the next day, remembering ORDERDB is a whole lot easier than remembering a long list of module names.
In addition, you can use PDM option 15 to compile the secondary source members. Since all the compiler parameters have been specified on the Header specification, a simple "15 enter" gets the source member compiled.
One other change can make compiling the entry module even easier. By adding the BNDDIR keyword to the entry module's Header specification, you can avoid using CRTRPGMOD (PDM option 15) for the entry module; instead, use CRTBNDRPG (PDM option 14) and avoid manually running the bind step (i.e., calling the CRTPGM command) entirely.
Here's what the entry module's Header specification looks like with the BNDDIR keyword added.
ORDENTRY | .....H OPTION(*NODEBUGIO : *SRCSTMT) .....H BNDDIR('ORDERBD') |
The new line in the Header specification above adds the BNDDIR keyword to the ORDENTRY source member. When using the BNDDIR keyword, make sure to specify the binding directory name in uppercase. If you enter lowercase letters, the compiler will not be able to locate the binding directory.
Bob Cozzi has been programming in RPG since 1978. Since then, he has written many articles and several books, including The Modern RPG Language --the most widely used RPG reference manual in the world. Bob is also a very popular speaker at industry events such as RPG World (www.rpgworld.com) and is the author of his own Web site (www.rpgiv.com) and of the RPG ToolKit (www.rpgiv.com/toolkit), an add-on library for RPG IV programmers.
LATEST COMMENTS
MC Press Online