Ok Thanks,

Won't that make me cut it loose from the dbc?


>No, you cannot open a table and not open structural CDX. You can make a copy of a table w/o CDX and work on it.
>
>>
>>Big legacy application 24/7, VFP7. I have a table (one of many) ~ 750000 records, with an integer pk, the next pk is stored in a separate table, retrieved and applied by an insert trigger in the dbc. Somewhere in the process it occasionally gains duplicate pks.
>>
>>Somewhere the error is lost. I can open the table, set order to any of the several tags, including the pk tag, with no complaints.
>>The index is getting corrupted somehow and I can't catch it, and of course I can't see the duplicates in the table.
>>
>>I can SQL select the duplicates to confirm they are present. I can delete tag all and browse the table to view the duplicates. Of course a reindex fails with a pk violation on that field due to the duplicates. Of course that is Sunday morning at 4:00.
>>
>>I can delete the tags, renumber the pk and rebuild the tags from metadata tables to work around the problem, but I have not been able to catch the actual event when the record gets the duplicate.
>>
>>I wanted to open the table without the cdx and without deleting the tags, so I could manually flag the records affected using other fields, then run the usual processes, and then still be able to find the records that had been duplicates after the renumbering. My thinking is to look for a pattern in the occurrence of the duplication, say, everytime Phoenix deletes one, Detroit uses that same key in its next append. Hence my question.
>>
>>Thanks,
>>
>>Al Allison