RE: Matching Records

["Joe Pluta" <joepluta@xxxxxxxxxxxxxxxxx>]

> From: David Williams
>
> Why not take the 15 minutes to write the matching
> record program that suits this situation rather than
> spend several hours figuring out all the possible way
> to make another less suited alternative work?  If I
> only had a dollar for each matching record program I
> wrote on the System/3 I could be retired by now.

Yep.  In this case, the entire program wouldn't be anything more than:

  If not *INMR or
    ((filea.field1 <> fileb.field1) or 
     (filea.field2 <> fileb.field2));
    Write PrintRec;
  Endif;

You'd need input specs to define the record identifying indicator for each
file (say 01 and 02) and to specify the match condition.  The program would
print one line for each record in only one file, and two lines for records
that are in both files both don't match.  Use 01 and 02 on the output spec
to specify which file is being printed.
 
If you were a little creative in your D-specs, you could just compare two
data structures instead of individual fields.

Joe