Practical RPG: Using QUSROBJD in Free-Format RPG

  • Smaller Small Medium Big Bigger
  • Default Helvetica Segoe Georgia Times

This article provides the details on how to use one of the most useful system management APIs, QUSROBJD.

Whether you're looking for unused objects, analyzing library differences, or checking how current your source is, QUSROBJD provides a wide range of valuable information—so much, in fact, that you might want to spend a little time up front to make it really easy to use in your programs. IBM provides predefined /COPY books to access their APIs, but this article will introduce you to a more modern approach that can be used for QUSROBJD and throughout the IBM i API universe.


QUSROBJD (officially known as the Retrieve Object Description API) provides a programmatic way to retrieve the information you would normally get through the DSPOBJD command. QUSROBJD is probably one of the first APIs I ever used, along with QWCCVTDT. I checked, and QUSROBJD has been around since V1R3 of the operating system, so it's not surprising that it's one of the first APIs I used.

QUSROBJD lets you retrieve nearly everything you need to know about an object. In fact, until the advent of ILE programming, there wasn't anything I needed to know about an object that I couldn't get from this API. A bit further on in this series of articles, we'll get to a discussion of the QBNLPGMI API and why we need it, but for now we can concentrate on QUSROBJD.

Like many IBM i APIs, QUSROBJD can return many different formats of data. The format of the returned data is specified as one of the parameters to the API. You can see the parameter list in Figure 1.

Practical RPG: Using QUSROBJD in Free-Format RPG - Figure 1

Figure 1: Here’s the parameter list for QUSROBJD from the IBM website.

As you can see, there aren't many parameters. QUSROBJD uses a relatively standard configuration shared by many IBM i APIs: a pointer to the receiver variable, its length, a format name that identifies the data to return, and one or more parameters to identify the object being queried. In this case, the values are the object library and name and the object type. The library and name are passed together in a single 20-character variable. This is very traditional in API programming; the object name is in the first 10 characters, and the object library is in the other 10. You'll notice two optional parameter groups. The word “group” is very loose; in this case, each group consists of a single parameter. I don't use the ASP parameter, but the error parameter is very important. For much more detail on it, please refer to my earlier article on API error handling.

So to clarify, the API can return several different configurations of information. Currently, QUSROBJD can return data in one of four layouts. Each has its own format identifier, and you specify that identifier as the third parameter when you call the API.

Doing Things the IBM Way

As I noted at the beginning of the article, IBM provides you with /COPY books to use to access the API. They're pretty barebones; they provide very simple definitions of the data structures that are geared toward programming in fixed-format RPG. They've got both RPG/400 and ILE RPG versions, but the ILE RPG uses fixed-format D-specs. Here is a comparison of the two styles (after removing the comments):

   I             'QUSROBJD'           C         QUSFDB

   IQUSFF     DS

   I                                   B   1   40QUSFFB

   I                                   B   5   80QUSFFC

   I                                      9 18 QUSFFD

   I                                    19 28 QUSFFF

Listing 1: This is the RPG/400 version of the QUSROBJD definition in QRPGSRC in QSYSINC.

     D QUSROBJD     C                'QUSROBJD'

     DQUSD0100       DS

     D QUSBRTN06          1     4I 0

     D QUSBAVL07          5     8I 0

     D QUSOBJN00          9     18

     D QUSOBJLN          19     28

Listing 2: This is the ILE RPG version of the QUSROBJD definition in QRPGLESRC in QSYSINC.

As you can see, it's a pretty low-tech conversion from one to the other. They updated the field names, and the numeric value changed from binary (B) to integer (I), but other than that it's still old-school RPG. Not that there's anything wrong with that, mind you; it meets the requirements of the lowest common denominator. But it's definitely not my preferred technique, especially the from and to positions for the field definitions. Something else becomes clear once you start looking at the other data structures, such as the one below for QUSD0200.

     DQUSD0200       DS

     D QUSBRTN07          1     4I 0

     D QUSBAVL08          5     8I 0

     D QUSOBJN01          9     18

     D QUSOBJLN00        19     28

Listing 3: The first lines of the OBJD0200 layout are nearly identical to the OBJD0100 definitions.

This is the definition for the OBJD0200 format, and you'll see that the first fields are the same as the first fields of the OBJD0100 format, except with slightly different names. Why is this? Well, the QUSROBJD API has a quirk in that every format is like the previous format but with more fields. In fact, in the definition on the IBM website, the first line of the definition for OBJD0200 literally says "everything from the OBJD0100 format." That being the case, you have to replicate all the fields for OBJD0100 in each of the four data structures. And since this is old-school RPG with no concept of qualified names, you have to have slightly different names for each field, hence the strange (and somewhat inconsistent) suffixes on the end of the repeated fields. As you can imagine, it's even worse in the /COPY books for RPG/400, with the six-character field name limitation.

In a fixed-format ILE RPG program, your code might look like this:

     C              CALL     QUSROBJD

     C              PARM                  QUSD0100

     C              PARM                  QUSD0100L    

     C              PARM     'OBJD0100'   QUSDFMT          10

     C              PARM                  OBJNAMLIB

     C              PARM                  OBJTYPE

     C              PARM                  ErrorDS

Listing 4: This would be a standard fixed-format call to the API.

There are a couple of additional definitions; the value QUSD0100L must be a four-byte integer (10I0, or int(10)) containing the length of the QUSD0100 structure, OBJNAMLIB must be a 20-character field with the object name and library as noted earlier, and OBJTYPE is a 10-character field containing the object type. ErrorDS needs to be a structure in one of the formats detailed in the API error handling article.

Freeing Your APIs

There's really no problem with the /COPY books that IBM provides. They're perfectly functional. But even after all that, you still have to figure out where the data is. And does a field name like QUSTD13 really indicate to you that it's the object's description? No, it doesn't. So instead, let me show you what I've done. Since the beginning of each data structure is exactly the same as the entire preceding data structure, I've created a set of templates. I've done one for each of the four structures, defining the fields unique to that structure. Here are the first two:

      dcl-ds T_OBJD_0100 template qualified;

      BytesRtn int(10);

      BytesAvl int(10);

      Object char(10);

      Library char(10);

      ObjectType char(10);

      LibFound char(10);

      ASP int(10);

      Owner char(10);

      Domain char(2);

      CreateTS13 char(13);

      ChangeTS13 char(13);


     dcl-ds T_OBJD_0200 template qualified;

      ExtendedAttr char(10);

      Description char(50);

      SrcFile char(10);

      SrcLib char(10);

      SrcMbr char(10);


Listing 5: These are the first two of the four templates for QUSROBJD.

The T_OBJD_0100 template contains all the fields for the OBJD0100 data structure. T_OBD_0200 contains all the fields for the OBJD0200 that are not in the OBJD0100 structure. The T_OBJD_0300 and T_OBJD_0400 templates follow the same pattern. You may notice that I do not use positioning for the subfields of the data structure. Instead, I define every spot in the data structure. This works as long as the data structure is entirely defined, which it is in the IBM /COPY books, even if they do occasionally resort to defining "reserved" sections. We can debate the pros and cons of sequential definitions versus positioned definitions another day, but sequential definitions are essential for the next part. Now I use those templates as building blocks for my actual structures:

    dcl-ds T_OBJD0100 qualified template;

        A0100 likeds(T_OBJD_0100);


     dcl-ds T_OBJD0200 qualified template;

        A0100 likeds(T_OBJD_0100);

        A0200 likeds(T_OBJD_0200);


     dcl-ds T_OBJD0300 qualified template;

        A0100 likeds(T_OBJD_0100);

        A0200 likeds(T_OBJD_0200);

        A0300 likeds(T_OBJD_0300);


     dcl-ds T_OBJD0400 qualified template;

        A0100 likeds(T_OBJD_0100);

        A0200 likeds(T_OBJD_0200);

        A0300 likeds(T_OBJD_0300);

        A0400 likeds(T_OBJD_0400);


Listing 6: Using the first template, I can now build the data structures.

So, as an example, my T_OBJD0300 structure contains three substructures, one for each set of fields. This is where qualified data structures become so much fun! Whenever I want to get at, say, the object owner, I simply access dsObject.A0100.Owner. It doesn't matter which data structure I used to define dsObject; the Owner subfield is always in the A0100 substructure. This is why I define even my highest-level structures as templates: that way, I can name my structures whatever I want in the calling program. Finally, I have the actual prototype here:

      // QUSROBJD - Retrieve object description

     dcl-pr QUSROBJD ExtPgm;

      iData char(32767) Options( *VarSize );

      iDataLen int(10) Const;

      iDataFmt char(8) Const;

      iObjQName char(20) Const;

      iObjType char(10) Const;

      ErrorDS char(32767) Options( *VarSize );


Listing 7: This is the prototype for the API itself.

Now that you have the infrastructure in place, calling the API becomes very simple. Let's take a situation: you want to write a procedure that will return the program type for a program (e.g., RPG, RPGLE, or CLLE). You can now write a procedure to do that with about 10 lines of code:

      dcl-proc getAttribute;

     dcl-pi *n char(10);

        iPgm char(10);

        iLib char(10);



     dcl-ds dsObject likeds(T_OBJD0200);

     dcl-ds apiError likeds(api_ErrorDS);


      QUSROBJD( dsObject: %size(dsObject): 'OBJD0200':

                iPgm+iLib: '*PGM': apiError);

     return dsObject.A0200.ExtendedAttr;


Listing 8: Using the free-format QUSROBJD prototype.

That's all there is to it. I'm glossing over a bit, such as error handling, but that's been covered in detail in the other article. What's important about this is how simple it is to call the API. I can easily call it to get additional information by simply using a different template (i.e., T_OBJD0400) and passing in a different format name. It's that easy.

Of course, the trick now is to create all those free-format API definitions for other APIs. I'm tempted to write a program that will extract that information from the IBM /COPY books. While the data structure field names are very cryptic, the comments are quite good, and I could probably extract meaningful field names from them.

It's something to think about. But until then, I hope you enjoyed this introduction to QUSROBJD and free-format APIs.