Hi Calvin

A couple of things to add to what others have said.

Under some circumstances, I have seen VFP try to add a duplicate record against a primary or candidate key. Generally it asks if this is OK, or do you wish to revert. Revert, and all is well, but the transaction will not be posted. On the other hand, OK saves the duplicate record, but it drops the Candidate/Primary setting on the tag. Not unreasonable - you can't have a candidate index that violates the rule. But the question means nothing to a user, and is time consuming to resolve. SDT will attempt to fix the problem, but will not be able to recreate the Candidate key until you have found and removed the duplicate and packed the table.

Re the header issue. I have found that while VFP automatically corrects the header problem, it only really fixes it if the file length is valid. I have had a couple of instances where the last record actually was one character short, visible through hexedit. Adding records failed until this was corrected.
VFP keeps trying to fix the header, doesn't quite succeed, and nothing is added to the table. But yes, the SDT repair function does fix the problem. The real problem, though, is the VFP did not report the problem, and you find out the hard way that it exists.

Geoff

>I have just recently converted a 10 year old FoxDOS program to VFP 6SP3. At this point approximately 400 users have converted over. Two very disconcerting things are happening. 1. Records are disappearing from files. 2. Primary Keys are being violated - that is, duplicate key fields are appearing in tables and the primary key index is being dropped.
>
>As to 1, I understand that the culprit is an incorrect record counter in the file header resulting in records being overwritten when new records are appended. This did happen in DOS from time to time but not with nearly the frequency it is happening now. As to 2, I did not think it was possible to insert a duplicate record into a key field. When this happens any screen using this table refuses to open which effectively shuts the program down.
>
>Two Questions - 1. Is the VFP file system stable? and 2. If it is, then why is this happening to me? By the way, I am running StoneField Database 5.1i and in my newest compile I am running FoxFix 5.0 in the background at startup to check file headers. I do not know if this is going to fix the problem.