>as you know, any update of the DBF header is going to require
>a lock of the file, and INSERT INTO is going to affect the DBF
>header. a flock imediately before the insert with a corresponding
>and an unlock immediately after should be all that is required.

I had the impression that INSERT INTO wasn't affecting the header. I think I know why I never had this error before. I didn't put the SET REPROCESS in this system. The default is 0 so if there is a collision, it failed right away by returning this message. I usually put it in all my system but for this one I forgot.