>Sylvain,
>
>>Oh and BTW, a workaround I've seen was to randomize the PK of a record before deleting it. It worked pretty well!
>
>Yes, they are what they are, workarrounds. Sure there are lot's of ways to solve the problem (which among the generated PKs) but this isn't constructive.
>
>The only problem with the xBase languages is that the deleted records still exist (in such way they still can be accessed and indexed). If this was not the case, the problem would not exist at all.
>
>Just think about it, would the problem still exist when deleted records simply are gone ..... ???
>
>It solves the problem with:
>- reusing PKs
>- The RI problem describe in my message to jim.
>- The need of filtered indexes
>- The problem of a non rushmore optimizible PK.
>
>My life would be so much easier if this feature was implemented in VFP 7.
>
>Regards,
>
>Walter,

It ain't gonna happen. It would break too much existing code.