Sylvain,

>>but note that the xBase solution is the wrong one.
>
>I see now why we don't agree. I don't think we ever will on this subject, so this conversation is pointless.
>
>Like Jim B., I never encountered problems with surrogate keys an often came across the same problem without them, so it's a justification all of itself to use them, thus the problem of deleted records are always solved automagically.

Yes, it works but (as i did describe in problem 3), not in all cases. Besides this I simply think it is wrong to use the Uniqueness violation as an argument to never reuse PKs. It is a fact that this problem only exist when not using filtered primary indexes. Outside the xBase language this problem does NOT exist !!

>Have a nice day! :)

Thanks,

Walter,