When the RPG IV development team designed built-in functions in the language, they omitted the functions that find the maximum and minimum among a group of values. This article presents these functions, plus two others I haven’t seen in any high-level language.

Often when comparing multiple numeric values, it’s useful to determine which value is the highest (maximum) and which is the lowest (minimum). Most programming languages offer a max() and a min() function precisely for this purpose, but RPG IV didn’t—until now. To demonstrate how useful these functions are, consider a typical sales analysis report: It determines which of the last 12 months provided the highest and lowest sales, information you need to know.

Functions such as max() and min() must be able to accept a variable number of parameters, since you may have to compare two numbers in one case, but 12 in another. Figure 1 shows the prototypes for four such functions: max(), maxid(), min(), and minid(). They have been defined with a maximum of 20 parameters, of which only the first and second are required. All others (third to 20th) bear OPTIONS(*NOPASS) to make them optional.

I have already described max() and min(). The other two functions don’t return the value of the maximum or minimum, but actually identify the value by its relative position within the parameter list. For instance, maxid() returns a 4 if the fourth parameter has the highest value. And minid() does the same for the minimum.

How They Work

All four functions have similar logic, so I’ll pick max() as an example of how the functions operate. Within subprocedure MAX (Figure 2), I define the procedure interface, which is a virtual duplicate of the prototype copied from MAXMIN@. I also declare a local variable maxval, which holds the maximum value found.

I initialize maxval to *LOVAL at the start of the C-specs. This makes maxval the potential minimum for all 20 parameters. Then, I compare it against nbr01, the first

parameter value. If nbr01 is greater than maxval, I update maxval accordingly. When this IF block ends, maxval will be either *LOVAL or equal to nbr01, whichever is greater.

Then, I repeat the process with the other 19 parameter values. When all 20 tests are performed, maxval will have the value of the maximum. Notice that, starting with the comparison against nbr03 (which is optional), the IFs ensure that the actual number of parameters received by max() includes the parameter in question. So, in the third IF block, I make sure that the function received is at least three parameters by comparing the built-in function %PARMS against 3. %PARMS always returns the number of parameters received by the current procedure.

How to Use It

Figure 3 shows a short RPG IV program that illustrates how to use the MAXMIN service program. All that remains after coding a /COPY directive to copy the prototypes is to use the functions.

Here, two of them are used: max() and minid(). In the first case, I determine that 30 is the highest of the numbers 10, 20, and 30; in the second case, the first parameter contains the lowest number.

Figure 3 illustrates something else. The receiving field, r, does not have to be defined to match the format of the function’s returned value. For example, the max() function returns the maximum value in a packed decimal field 30 digits long. Field r, however, has only seven digits but contains two decimal positions; yet the function works, and the receiving field is updated correctly.

One last item: All four functions accept only whole numbers as parameters. If you need to compare numbers that may have decimal positions, you’ll have to convert them to whole numbers first. Fortunately, you can do this with expressions already embedded within the function itself. For instance, if you want to find the minimum between field a and b (each having two decimal positions), you can code the min() function as follows:

min(a*100:b*100)

You guarantee that both numbers are whole numbers because each is multiplied by
100.