The solutions mentioned so far indeed do solve your functional problems of
getting rid of all the master entries. However, they do not address the
associated performance problems you cited - since you will experience
migrating secondaries each time you delete a master entry which is a primary
entry and has at least one secondary.
 
So......., to both complete the task properly AND have it done EFFICIENTLY,
one way to accomplish these is to:
 
loop1:
        read next master entry at end go to part2.
                is it a secondary (words 5-6, relative to 1, of status array =
0)
                if yes
                   delete it.
                go to loop1.
 
at this point, you will have only master entries left.
 
part2:
        dbclose mode 3 (rewind dataset)
loop2:
        read next master entry at end JOB WELL DONE
        delete
        go to loop2:
 
 
Please forgive my lousy pseudo-code.
 
 
 
****************************************************
Gilles Schipper
GSA Inc.
Tel: 905/889-3000
Fax: 905/889-3001
****************************************************
 
 
 
--
 
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
             Magic Online Services Toronto Inc.   (416) 591-6490
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-