By Shawn M. Gordon

I’ve recently had to deal with sorting at various levels and got into some conversations about the best approach. There are a couple of issues with sorting. First, is it just a small table that you want to sort in memory? If it is, then a bubble or shell sort might be the way to go, in order to avoid writing the data to a file. I will cover these techniques in a later column.

For sorting files you have several options. You can call the sort intrinsics directly, which is nasty. You can take advantage of the FD and SD and call the SORT verb in your code, or use the variation of the SORT verb with INPUT/OUTPUT procedures using RELEASE and RETURN. I am going to cover the last technique, because it is the one that is most useful and least used.

I remember when I first saw RELEASE/RETURN, a friend of mine was learning COBOL and he was asking me about it. I immediately responded that there was no such thing, and what the heck was his teacher telling him? I finally decided to look it up — and boy, was my face red. But I didn’t feel too bad, because the description in the manual is so awful. Check this out:

“The RETURN statement can be used in an output procedure of a SORT or MERGE statement. It cannot be used in any other type of procedure. For more information on the RETURN statement refer to Chapter 12, “SORT/MERGE Operations”.

“The RELEASE statement can be used in an input procedure of a SORT statement to transfer records from your program to the initial phase of the sort operation. For more information on the RELEASE statement refer to Chapter 12, 'SORT/MERGE Operations.' ”

Chapter 12: SORT/MERGE Operations

“The sort/merge capabilities of HP COBOL II allow you to sort one or more files of records, or to combine two or more identically ordered files of records one or more times within a given execution of a COBOL program.

“Additionally, you have the ability to specially process individual records by input or output procedures that are part of the sort or merge operation. For a sort operation, this special processing may be applied before as well as after the records have been sorted. For a merge operation, this processing may be applied after the records have been combined. The RELEASE and RETURN statements are used in these input procedures to release or return records that are to be, or have been sorted or merged.”

Now if that explains how to do it, then I’m impressed with whoever figures it out. I finally found out how to do this from a fine programmer by the name of Alan Oshiro at a job in the late 80s, who I have not been able to track down for many years now. In any case, you all get to take advantage of that time Alan spent with me. Figure 1 below shows how to use RELEASE and RETURN.

Now the code fragment in Figure 1 is a bit light on details, but should hopefully give you enough to get started. I suggest that you play around with this technique a bit and make sure you understand how it is working. Like I said, it is (or at least the last time I checked) just slightly faster than using the standard SORT verb in COBOL.

As always, I welcome your input to this column. Earn yourself a free 3000 Always Online hat for tips you submit, either to the NewsWire or to me at shawn@smga3000.com.