re: alphabetic header file

[Peter Coffin <phcoffin@xxxxxxxxxx>]

Booth writes:

> Lets say I have a name and address header file where the records include
> these five names:

> Adams Brewery, Inc.
> J. A. Adams & Company, Inc.
> Bruce Adams
> Bruce Adams, Jr.
> Adam & Bruce Bagels & Buns

> Now, I want one of the logical files over that physical file to provide
> alphabetic order, using phone book rules of alphabetization.  What
> algorithms make sense and are useful and efficient?

First, either separate out the two kinds of information you have into
separate files, or define that you're looking at entities and not people.
EG: pretend Bruce Adams is Bruce Adams, Inc. and sort accordingly.

If Bruce is to be an individual, then do this:

Enterprise                  Contact           Address        (etc.)
--------------------------- ----------------- -------------- ------------
Adams Brewery, Inc.                           123 Main St.
J. A. Adams & Company, Inc.                   456 State St.
                            Adams, Bruce      789 Oak St.
                            Adams, Bruce Jr.  819 Oak St.
Adam & Bruce Bagels & Buns                    1400 Maple St.

and populate the missing fields as necessary.

Of course, if the file is already mixed up as shown in your example,
you'll have to send someone through the file to decide whether a
record is an individual or an enterprise and put some sort of an
indicator in a field in the the mixed file so you can then process
it out into a more final form. In that case, redefining Bruce Adams
as Bruce Adams (Inc.) may be your quicker solution.

Peter H. Coffin
T/L 665-6298 (414)223-6298
phcoffin@us.ibm.com
+---
| This is the Midrange System Mailing List!
| To submit a new message, send your mail to MIDRANGE-L@midrange.com.
| To unsubscribe from this list send email to MIDRANGE-L-UNSUB@midrange.com.
| Questions should be directed to the list owner/operator: david@midrange.com
+---