Hi Cecil

Good that you found the bug. Consider there should have been no reason to do a seek. If a record was appended, either don't move the record pointer, or if any procedures are called after this append, ensure that such procedures push and pop the record #.

>There is no filter. I didn't go in and move the record pointer in my debugging, but I could, but I seem to have resolved the problem already.
>
>The problem was that since the seek wasn't working, I was getting duplicate records. The way the person who coded this, prior to my time here, was that he first created a record by APPENDING then went back and sought the record when he was ready to copy over some field data to the same record. Unfortunately, his search expression was different than the Index TAG's search expression, which left out one small part of the expression. For example:
>
>The search expresison was: PO_MAIN
>The Index TAG's expression was: PO_MAIN + PO_SUFFIX
>
>I believe that the SET EXACT setting must have been set differently which might explain why this worked, or seemed to work for a long time. Perhaps we set the EXACT setting somewhere in the code differently than what this search needed and this could explain why the search worked for so long and all of a sudden it didn't work.
>
>Final result was that I made the search and the Index TAG expressions match precisely. No more duplicate records are created and it all works like it should now.
>
>Nice story, huh? Fortunately, this story is true and has a good ending. But, unfortunately, it has taken a week, off and on, to find this bug.